Podcasts about tpu

Et si l'avenir de l'intelligence artificielle ne se jouait plus au sol, mais à 500 kilomètres au-dessus de nos têtes ?C'est le pari fou, mais très sérieux, de Google et la société Planet avec le projet Suncatcher. L'objectif est simple : construire les premiers centres de données orbitaux.Les deux géants viennent d'annoncer une accélération majeure avec le lancement prévu de deux satellites de démonstration d'ici début 2027.Une alimentation solaire quasi continueLe premier pilier de cette stratégie repose sur la résolution d'une équation énergétique devenue critique sur Terre.Aujourd'hui, les centres de données IA saturent les réseaux électriques et posent des problèmes de refroidissement colossaux.En plaçant ces serveurs en orbite héliosynchrone, Google et Planet visent une alimentation solaire quasi continue.Concrètement, ces satellites qui serviraient de "fermes de calcul" déploieront des panneaux solaires XXL pour alimenter des puces TPU, les processeurs de Google optimisés pour l'IA.Mais le vrai défi technique reste la dissipation de la chaleur dans le vide spatial et la protection des composants contre les radiations. C'est tout l'enjeu des tests de 2027. Il s'agit de prouver qu'on peut faire tourner un cluster de calcul intensif dans l'hostilité de l'espace.Un cluster spatial volantEnsuite, le projet Suncatcher inaugure une architecture réseau d'un genre nouveau : le cluster spatial volant.On ne parle pas de satellites isolés, mais de grappes de machines situées à moins de 200 mètres les unes des autres, reliées par des liaisons laser à très haut débit.Cela préfigure un cloud hybride totalement indépendant des infrastructures terrestres et capable de traiter les données directement dans l'espace.Planet utilise ici son expérience unique, ayant déjà mis en orbite plus de 600 satellites, pour industrialiser ce que Google appelle un "système de cluster à grande échelle".Vers le développement industrielEnfin, cette alliance marque un tournant concurrentiel majeur dans la course à l'espace.Si Jeff Bezos et Elon Musk ont déjà évoqué l'idée de data centers spatiaux, Google et Planet sont les premiers à passer concrètement en phase de recherche et développement industrielle.Le PDG de Planet, Will Marshall, l'affirme : nous ne sommes qu'à quelques années du point de bascule économique où l'espace deviendra moins cher que la Terre pour le calcul intensif.Avec la baisse drastique des coûts de lancement, l'infrastructure spatiale devient un levier stratégique pour la puissance de calcul.Le ZD Tech est sur toutes les plateformes de podcast ! Abonnez-vous !Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.

From rewriting Google's search stack in the early 2000s to reviving sparse trillion-parameter models and co-designing TPUs with frontier ML research, Jeff Dean has quietly shaped nearly every layer of the modern AI stack. As Chief AI Scientist at Google and a driving force behind Gemini, Jeff has lived through multiple scaling revolutions from CPUs and sharded indices to multimodal models that reason across text, video, and code.Jeff joins us to unpack what it really means to “own the Pareto frontier,” why distillation is the engine behind every Flash model breakthrough, how energy (in picojoules) not FLOPs is becoming the true bottleneck, what it was like leading the charge to unify all of Google's AI teams, and why the next leap won't come from bigger context windows alone, but from systems that give the illusion of attending to trillions of tokens.We discuss:* Jeff's early neural net thesis in 1990: parallel training before it was cool, why he believed scaling would win decades early, and the “bigger model, more data, better results” mantra that held for 15 years* The evolution of Google Search: sharding, moving the entire index into memory in 2001, softening query semantics pre-LLMs, and why retrieval pipelines already resemble modern LLM systems* Pareto frontier strategy: why you need both frontier “Pro” models and low-latency “Flash” models, and how distillation lets smaller models surpass prior generations* Distillation deep dive: ensembles → compression → logits as soft supervision, and why you need the biggest model to make the smallest one good* Latency as a first-class objective: why 10–50x lower latency changes UX entirely, and how future reasoning workloads will demand 10,000 tokens/sec* Energy-based thinking: picojoules per bit, why moving data costs 1000x more than a multiply, batching through the lens of energy, and speculative decoding as amortization* TPU co-design: predicting ML workloads 2–6 years out, speculative hardware features, precision reduction, sparsity, and the constant feedback loop between model architecture and silicon* Sparse models and “outrageously large” networks: trillions of parameters with 1–5% activation, and why sparsity was always the right abstraction* Unified vs. specialized models: abandoning symbolic systems, why general multimodal models tend to dominate vertical silos, and when vertical fine-tuning still makes sense* Long context and the illusion of scale: beyond needle-in-a-haystack benchmarks toward systems that narrow trillions of tokens to 117 relevant documents* Personalized AI: attending to your emails, photos, and documents (with permission), and why retrieval + reasoning will unlock deeply personal assistants* Coding agents: 50 AI interns, crisp specifications as a new core skill, and how ultra-low latency will reshape human–agent collaboration* Why ideas still matter: transformers, sparsity, RL, hardware, systems — scaling wasn't blind; the pieces had to multiply togetherShow Notes:* Gemma 3 Paper* Gemma 3* Gemini 2.5 Report* Jeff Dean's “Software Engineering Advice fromBuilding Large-Scale Distributed Systems” Presentation (with Back of the Envelope Calculations)* Latency Numbers Every Programmer Should Know by Jeff Dean* The Jeff Dean Facts* Jeff Dean Google Bio* Jeff Dean on “Important AI Trends” @Stanford AI Club* Jeff Dean & Noam Shazeer — 25 years at Google (Dwarkesh)—Jeff Dean* LinkedIn: https://www.linkedin.com/in/jeff-dean-8b212555* X: https://x.com/jeffdeanGoogle* https://google.com* https://deepmind.googleFull Video EpisodeTimestamps00:00:04 — Introduction: Alessio & Swyx welcome Jeff Dean, chief AI scientist at Google, to the Latent Space podcast00:00:30 — Owning the Pareto Frontier & balancing frontier vs low-latency models00:01:31 — Frontier models vs Flash models + role of distillation00:03:52 — History of distillation and its original motivation00:05:09 — Distillation's role in modern model scaling00:07:02 — Model hierarchy (Flash, Pro, Ultra) and distillation sources00:07:46 — Flash model economics & wide deployment00:08:10 — Latency importance for complex tasks00:09:19 — Saturation of some tasks and future frontier tasks00:11:26 — On benchmarks, public vs internal00:12:53 — Example long-context benchmarks & limitations00:15:01 — Long-context goals: attending to trillions of tokens00:16:26 — Realistic use cases beyond pure language00:18:04 — Multimodal reasoning and non-text modalities00:19:05 — Importance of vision & motion modalities00:20:11 — Video understanding example (extracting structured info)00:20:47 — Search ranking analogy for LLM retrieval00:23:08 — LLM representations vs keyword search00:24:06 — Early Google search evolution & in-memory index00:26:47 — Design principles for scalable systems00:28:55 — Real-time index updates & recrawl strategies00:30:06 — Classic “Latency numbers every programmer should know”00:32:09 — Cost of memory vs compute and energy emphasis00:34:33 — TPUs & hardware trade-offs for serving models00:35:57 — TPU design decisions & co-design with ML00:38:06 — Adapting model architecture to hardware00:39:50 — Alternatives: energy-based models, speculative decoding00:42:21 — Open research directions: complex workflows, RL00:44:56 — Non-verifiable RL domains & model evaluation00:46:13 — Transition away from symbolic systems toward unified LLMs00:47:59 — Unified models vs specialized ones00:50:38 — Knowledge vs reasoning & retrieval + reasoning00:52:24 — Vertical model specialization & modules00:55:21 — Token count considerations for vertical domains00:56:09 — Low resource languages & contextual learning00:59:22 — Origins: Dean's early neural network work01:10:07 — AI for coding & human–model interaction styles01:15:52 — Importance of crisp specification for coding agents01:19:23 — Prediction: personalized models & state retrieval01:22:36 — Token-per-second targets (10k+) and reasoning throughput01:23:20 — Episode conclusion and thanksTranscriptAlessio Fanelli [00:00:04]: Hey everyone, welcome to the Latent Space podcast. This is Alessio, founder of Kernel Labs, and I'm joined by Swyx, editor of Latent Space. Shawn Wang [00:00:11]: Hello, hello. We're here in the studio with Jeff Dean, chief AI scientist at Google. Welcome. Thanks for having me. It's a bit surreal to have you in the studio. I've watched so many of your talks, and obviously your career has been super legendary. So, I mean, congrats. I think the first thing must be said, congrats on owning the Pareto Frontier.Jeff Dean [00:00:30]: Thank you, thank you. Pareto Frontiers are good. It's good to be out there.Shawn Wang [00:00:34]: Yeah, I mean, I think it's a combination of both. You have to own the Pareto Frontier. You have to have like frontier capability, but also efficiency, and then offer that range of models that people like to use. And, you know, some part of this was started because of your hardware work. Some part of that is your model work, and I'm sure there's lots of secret sauce that you guys have worked on cumulatively. But, like, it's really impressive to see it all come together in, like, this slittily advanced.Jeff Dean [00:01:04]: Yeah, yeah. I mean, I think, as you say, it's not just one thing. It's like a whole bunch of things up and down the stack. And, you know, all of those really combine to help make UNOS able to make highly capable large models, as well as, you know, software techniques to get those large model capabilities into much smaller, lighter weight models that are, you know, much more cost effective and lower latency, but still, you know, quite capable for their size. Yeah.Alessio Fanelli [00:01:31]: How much pressure do you have on, like, having the lower bound of the Pareto Frontier, too? I think, like, the new labs are always trying to push the top performance frontier because they need to raise more money and all of that. And you guys have billions of users. And I think initially when you worked on the CPU, you were thinking about, you know, if everybody that used Google, we use the voice model for, like, three minutes a day, they were like, you need to double your CPU number. Like, what's that discussion today at Google? Like, how do you prioritize frontier versus, like, we have to do this? How do we actually need to deploy it if we build it?Jeff Dean [00:02:03]: Yeah, I mean, I think we always want to have models that are at the frontier or pushing the frontier because I think that's where you see what capabilities now exist that didn't exist at the sort of slightly less capable last year's version or last six months ago version. At the same time, you know, we know those are going to be really useful for a bunch of use cases, but they're going to be a bit slower and a bit more expensive than people might like for a bunch of other broader models. So I think what we want to do is always have kind of a highly capable sort of affordable model that enables a whole bunch of, you know, lower latency use cases. People can use them for agentic coding much more readily and then have the high-end, you know, frontier model that is really useful for, you know, deep reasoning, you know, solving really complicated math problems, those kinds of things. And it's not that. One or the other is useful. They're both useful. So I think we'd like to do both. And also, you know, through distillation, which is a key technique for making the smaller models more capable, you know, you have to have the frontier model in order to then distill it into your smaller model. So it's not like an either or choice. You sort of need that in order to actually get a highly capable, more modest size model. Yeah.Alessio Fanelli [00:03:24]: I mean, you and Jeffrey came up with the solution in 2014.Jeff Dean [00:03:28]: Don't forget, L'Oreal Vinyls as well. Yeah, yeah.Alessio Fanelli [00:03:30]: A long time ago. But like, I'm curious how you think about the cycle of these ideas, even like, you know, sparse models and, you know, how do you reevaluate them? How do you think about in the next generation of model, what is worth revisiting? Like, yeah, they're just kind of like, you know, you worked on so many ideas that end up being influential, but like in the moment, they might not feel that way necessarily. Yeah.Jeff Dean [00:03:52]: I mean, I think distillation was originally motivated because we were seeing that we had a very large image data set at the time, you know, 300 million images that we could train on. And we were seeing that if you create specialists for different subsets of those image categories, you know, this one's going to be really good at sort of mammals, and this one's going to be really good at sort of indoor room scenes or whatever, and you can cluster those categories and train on an enriched stream of data after you do pre-training on a much broader set of images. You get much better performance. If you then treat that whole set of maybe 50 models you've trained as a large ensemble, but that's not a very practical thing to serve, right? So distillation really came about from the idea of, okay, what if we want to actually serve that and train all these independent sort of expert models and then squish it into something that actually fits in a form factor that you can actually serve? And that's, you know, not that different from what we're doing today. You know, often today we're instead of having an ensemble of 50 models. We're having a much larger scale model that we then distill into a much smaller scale model.Shawn Wang [00:05:09]: Yeah. A part of me also wonders if distillation also has a story with the RL revolution. So let me maybe try to articulate what I mean by that, which is you can, RL basically spikes models in a certain part of the distribution. And then you have to sort of, well, you can spike models, but usually sometimes... It might be lossy in other areas and it's kind of like an uneven technique, but you can probably distill it back and you can, I think that the sort of general dream is to be able to advance capabilities without regressing on anything else. And I think like that, that whole capability merging without loss, I feel like it's like, you know, some part of that should be a distillation process, but I can't quite articulate it. I haven't seen much papers about it.Jeff Dean [00:06:01]: Yeah, I mean, I tend to think of one of the key advantages of distillation is that you can have a much smaller model and you can have a very large, you know, training data set and you can get utility out of making many passes over that data set because you're now getting the logits from the much larger model in order to sort of coax the right behavior out of the smaller model that you wouldn't otherwise get with just the hard labels. And so, you know, I think that's what we've observed. Is you can get, you know, very close to your largest model performance with distillation approaches. And that seems to be, you know, a nice sweet spot for a lot of people because it enables us to kind of, for multiple Gemini generations now, we've been able to make the sort of flash version of the next generation as good or even substantially better than the previous generations pro. And I think we're going to keep trying to do that because that seems like a good trend to follow.Shawn Wang [00:07:02]: So, Dara asked, so it was the original map was Flash Pro and Ultra. Are you just sitting on Ultra and distilling from that? Is that like the mother load?Jeff Dean [00:07:12]: I mean, we have a lot of different kinds of models. Some are internal ones that are not necessarily meant to be released or served. Some are, you know, our pro scale model and we can distill from that as well into our Flash scale model. So I think, you know, it's an important set of capabilities to have and also inference time scaling. It can also be a useful thing to improve the capabilities of the model.Shawn Wang [00:07:35]: And yeah, yeah, cool. Yeah. And obviously, I think the economy of Flash is what led to the total dominance. I think the latest number is like 50 trillion tokens. I don't know. I mean, obviously, it's changing every day.Jeff Dean [00:07:46]: Yeah, yeah. But, you know, by market share, hopefully up.Shawn Wang [00:07:50]: No, I mean, there's no I mean, there's just the economics wise, like because Flash is so economical, like you can use it for everything. Like it's in Gmail now. It's in YouTube. Like it's yeah. It's in everything.Jeff Dean [00:08:02]: We're using it more in our search products of various AI mode reviews.Shawn Wang [00:08:05]: Oh, my God. Flash past the AI mode. Oh, my God. Yeah, that's yeah, I didn't even think about that.Jeff Dean [00:08:10]: I mean, I think one of the things that is quite nice about the Flash model is not only is it more affordable, it's also a lower latency. And I think latency is actually a pretty important characteristic for these models because we're going to want models to do much more complicated things that are going to involve, you know, generating many more tokens from when you ask the model to do so. So, you know, if you're going to ask the model to do something until it actually finishes what you ask it to do, because you're going to ask now, not just write me a for loop, but like write me a whole software package to do X or Y or Z. And so having low latency systems that can do that seems really important. And Flash is one direction, one way of doing that. You know, obviously our hardware platforms enable a bunch of interesting aspects of our, you know, serving stack as well, like TPUs, the interconnect between. Chips on the TPUs is actually quite, quite high performance and quite amenable to, for example, long context kind of attention operations, you know, having sparse models with lots of experts. These kinds of things really, really matter a lot in terms of how do you make them servable at scale.Alessio Fanelli [00:09:19]: Yeah. Does it feel like there's some breaking point for like the proto Flash distillation, kind of like one generation delayed? I almost think about almost like the capability as a. In certain tasks, like the pro model today is a saturated, some sort of task. So next generation, that same task will be saturated at the Flash price point. And I think for most of the things that people use models for at some point, the Flash model in two generation will be able to do basically everything. And how do you make it economical to like keep pushing the pro frontier when a lot of the population will be okay with the Flash model? I'm curious how you think about that.Jeff Dean [00:09:59]: I mean, I think that's true. If your distribution of what people are asking people, the models to do is stationary, right? But I think what often happens is as the models become more capable, people ask them to do more, right? So, I mean, I think this happens in my own usage. Like I used to try our models a year ago for some sort of coding task, and it was okay at some simpler things, but wouldn't do work very well for more complicated things. And since then, we've improved dramatically on the more complicated coding tasks. And now I'll ask it to do much more complicated things. And I think that's true, not just of coding, but of, you know, now, you know, can you analyze all the, you know, renewable energy deployments in the world and give me a report on solar panel deployment or whatever. That's a very complicated, you know, more complicated task than people would have asked a year ago. And so you are going to want more capable models to push the frontier in the absence of what people ask the models to do. And that also then gives us. Insight into, okay, where does the, where do things break down? How can we improve the model in these, these particular areas, uh, in order to sort of, um, make the next generation even better.Alessio Fanelli [00:11:11]: Yeah. Are there any benchmarks or like test sets they use internally? Because it's almost like the same benchmarks get reported every time. And it's like, all right, it's like 99 instead of 97. Like, how do you have to keep pushing the team internally to it? Or like, this is what we're building towards. Yeah.Jeff Dean [00:11:26]: I mean, I think. Benchmarks, particularly external ones that are publicly available. Have their utility, but they often kind of have a lifespan of utility where they're introduced and maybe they're quite hard for current models. You know, I, I like to think of the best kinds of benchmarks are ones where the initial scores are like 10 to 20 or 30%, maybe, but not higher. And then you can sort of work on improving that capability for, uh, whatever it is, the benchmark is trying to assess and get it up to like 80, 90%, whatever. I, I think once it hits kind of 95% or something, you get very diminishing returns from really focusing on that benchmark, cuz it's sort of, it's either the case that you've now achieved that capability, or there's also the issue of leakage in public data or very related kind of data being, being in your training data. Um, so we have a bunch of held out internal benchmarks that we really look at where we know that wasn't represented in the training data at all. There are capabilities that we want the model to have. Um, yeah. Yeah. Um, that it doesn't have now, and then we can work on, you know, assessing, you know, how do we make the model better at these kinds of things? Is it, we need different kind of data to train on that's more specialized for this particular kind of task. Do we need, um, you know, a bunch of, uh, you know, architectural improvements or some sort of, uh, model capability improvements, you know, what would help make that better?Shawn Wang [00:12:53]: Is there, is there such an example that you, uh, a benchmark inspired in architectural improvement? Like, uh, I'm just kind of. Jumping on that because you just.Jeff Dean [00:13:02]: Uh, I mean, I think some of the long context capability of the, of the Gemini models that came, I guess, first in 1.5 really were about looking at, okay, we want to have, um, you know,Shawn Wang [00:13:15]: immediately everyone jumped to like completely green charts of like, everyone had, I was like, how did everyone crack this at the same time? Right. Yeah. Yeah.Jeff Dean [00:13:23]: I mean, I think, um, and once you're set, I mean, as you say that needed single needle and a half. Hey, stack benchmark is really saturated for at least context links up to 1, 2 and K or something. Don't actually have, you know, much larger than 1, 2 and 8 K these days or two or something. We're trying to push the frontier of 1 million or 2 million context, which is good because I think there are a lot of use cases where. Yeah. You know, putting a thousand pages of text or putting, you know, multiple hour long videos and the context and then actually being able to make use of that as useful. Try to, to explore the über graduation are fairly large. But the single needle in a haystack benchmark is sort of saturated. So you really want more complicated, sort of multi-needle or more realistic, take all this content and produce this kind of answer from a long context that sort of better assesses what it is people really want to do with long context. Which is not just, you know, can you tell me the product number for this particular thing?Shawn Wang [00:14:31]: Yeah, it's retrieval. It's retrieval within machine learning. It's interesting because I think the more meta level I'm trying to operate at here is you have a benchmark. You're like, okay, I see the architectural thing I need to do in order to go fix that. But should you do it? Because sometimes that's an inductive bias, basically. It's what Jason Wei, who used to work at Google, would say. Exactly the kind of thing. Yeah, you're going to win. Short term. Longer term, I don't know if that's going to scale. You might have to undo that.Jeff Dean [00:15:01]: I mean, I like to sort of not focus on exactly what solution we're going to derive, but what capability would you want? And I think we're very convinced that, you know, long context is useful, but it's way too short today. Right? Like, I think what you would really want is, can I attend to the internet while I answer my question? Right? But that's not going to happen. I think that's going to be solved by purely scaling the existing solutions, which are quadratic. So a million tokens kind of pushes what you can do. You're not going to do that to a trillion tokens, let alone, you know, a billion tokens, let alone a trillion. But I think if you could give the illusion that you can attend to trillions of tokens, that would be amazing. You'd find all kinds of uses for that. You would have attend to the internet. You could attend to the pixels of YouTube and the sort of deeper representations that we can find. You could attend to the form for a single video, but across many videos, you know, on a personal Gemini level, you could attend to all of your personal state with your permission. So like your emails, your photos, your docs, your plane tickets you have. I think that would be really, really useful. And the question is, how do you get algorithmic improvements and system level improvements that get you to something where you actually can attend to trillions of tokens? Right. In a meaningful way. Yeah.Shawn Wang [00:16:26]: But by the way, I think I did some math and it's like, if you spoke all day, every day for eight hours a day, you only generate a maximum of like a hundred K tokens, which like very comfortably fits.Jeff Dean [00:16:38]: Right. But if you then say, okay, I want to be able to understand everything people are putting on videos.Shawn Wang [00:16:46]: Well, also, I think that the classic example is you start going beyond language into like proteins and whatever else is extremely information dense. Yeah. Yeah.Jeff Dean [00:16:55]: I mean, I think one of the things about Gemini's multimodal aspects is we've always wanted it to be multimodal from the start. And so, you know, that sometimes to people means text and images and video sort of human-like and audio, audio, human-like modalities. But I think it's also really useful to have Gemini know about non-human modalities. Yeah. Like LIDAR sensor data from. Yes. Say, Waymo vehicles or. Like robots or, you know, various kinds of health modalities, x-rays and MRIs and imaging and genomics information. And I think there's probably hundreds of modalities of data where you'd like the model to be able to at least be exposed to the fact that this is an interesting modality and has certain meaning in the world. Where even if you haven't trained on all the LIDAR data or MRI data, you could have, because maybe that's not, you know, it doesn't make sense in terms of trade-offs of. You know, what you include in your main pre-training data mix, at least including a little bit of it is actually quite useful. Yeah. Because it sort of tempts the model that this is a thing.Shawn Wang [00:18:04]: Yeah. Do you believe, I mean, since we're on this topic and something I just get to ask you all the questions I always wanted to ask, which is fantastic. Like, are there some king modalities, like modalities that supersede all the other modalities? So a simple example was Vision can, on a pixel level, encode text. And DeepSeq had this DeepSeq CR paper that did that. Vision. And Vision has also been shown to maybe incorporate audio because you can do audio spectrograms and that's, that's also like a Vision capable thing. Like, so, so maybe Vision is just the king modality and like. Yeah.Jeff Dean [00:18:36]: I mean, Vision and Motion are quite important things, right? Motion. Well, like video as opposed to static images, because I mean, there's a reason evolution has evolved eyes like 23 independent ways, because it's such a useful capability for sensing the world around you, which is really what we want these models to be. So I think the only thing that we can be able to do is interpret the things we're seeing or the things we're paying attention to and then help us in using that information to do things. Yeah.Shawn Wang [00:19:05]: I think motion, you know, I still want to shout out, I think Gemini, still the only native video understanding model that's out there. So I use it for YouTube all the time. Nice.Jeff Dean [00:19:15]: Yeah. Yeah. I mean, it's actually, I think people kind of are not necessarily aware of what the Gemini models can actually do. Yeah. Like I have an example I've used in one of my talks. It had like, it was like a YouTube highlight video of 18 memorable sports moments across the last 20 years or something. So it has like Michael Jordan hitting some jump shot at the end of the finals and, you know, some soccer goals and things like that. And you can literally just give it the video and say, can you please make me a table of what all these different events are? What when the date is when they happened? And a short description. And so you get like now an 18 row table of that information extracted from the video, which is, you know, not something most people think of as like a turn video into sequel like table.Alessio Fanelli [00:20:11]: Has there been any discussion inside of Google of like, you mentioned tending to the whole internet, right? Google, it's almost built because a human cannot tend to the whole internet and you need some sort of ranking to find what you need. Yep. That ranking is like much different for an LLM because you can expect a person to look at maybe the first five, six links in a Google search versus for an LLM. Should you expect to have 20 links that are highly relevant? Like how do you internally figure out, you know, how do we build the AI mode that is like maybe like much broader search and span versus like the more human one? Yeah.Jeff Dean [00:20:47]: I mean, I think even pre-language model based work, you know, our ranking systems would be built to start. I mean, I think even pre-language model based work, you know, our ranking systems would be built to start. With a giant number of web pages in our index, many of them are not relevant. So you identify a subset of them that are relevant with very lightweight kinds of methods. You know, you're down to like 30,000 documents or something. And then you gradually refine that to apply more and more sophisticated algorithms and more and more sophisticated sort of signals of various kinds in order to get down to ultimately what you show, which is, you know, the final 10 results or, you know, 10 results plus. Other kinds of information. And I think an LLM based system is not going to be that dissimilar, right? You're going to attend to trillions of tokens, but you're going to want to identify, you know, what are the 30,000 ish documents that are with the, you know, maybe 30 million interesting tokens. And then how do you go from that into what are the 117 documents I really should be paying attention to in order to carry out the tasks that the user has asked? And I think, you know, you can imagine systems where you have, you know, a lot of highly parallel processing to identify those initial 30,000 candidates, maybe with very lightweight kinds of models. Then you have some system that sort of helps you narrow down from 30,000 to the 117 with maybe a little bit more sophisticated model or set of models. And then maybe the final model is the thing that looks. So the 117 things that might be your most capable model. So I think it has to, it's going to be some system like that, that is really enables you to give the illusion of attending to trillions of tokens. Sort of the way Google search gives you, you know, not the illusion, but you are searching the internet, but you're finding, you know, a very small subset of things that are, that are relevant.Shawn Wang [00:22:47]: Yeah. I often tell a lot of people that are not steeped in like Google search history that, well, you know, like Bert was. Like he was like basically immediately inside of Google search and that improves results a lot, right? Like I don't, I don't have any numbers off the top of my head, but like, I'm sure you guys, that's obviously the most important numbers to Google. Yeah.Jeff Dean [00:23:08]: I mean, I think going to an LLM based representation of text and words and so on enables you to get out of the explicit hard notion of, of particular words having to be on the page, but really getting at the notion of this topic of this page or this page. Paragraph is highly relevant to this query. Yeah.Shawn Wang [00:23:28]: I don't think people understand how much LLMs have taken over all these very high traffic system, very high traffic. Yeah. Like it's Google, it's YouTube. YouTube has this like semantics ID thing where it's just like every token or every item in the vocab is a YouTube video or something that predicts the video using a code book, which is absurd to me for YouTube size.Jeff Dean [00:23:50]: And then most recently GROK also for, for XAI, which is like, yeah. I mean, I'll call out even before LLMs were used extensively in search, we put a lot of emphasis on softening the notion of what the user actually entered into the query.Shawn Wang [00:24:06]: So do you have like a history of like, what's the progression? Oh yeah.Jeff Dean [00:24:09]: I mean, I actually gave a talk in, uh, I guess, uh, web search and data mining conference in 2009, uh, where we never actually published any papers about the origins of Google search, uh, sort of, but we went through sort of four or five or six. generations, four or five or six generations of, uh, redesigning of the search and retrieval system, uh, from about 1999 through 2004 or five. And that talk is really about that evolution. And one of the things that really happened in 2001 was we were sort of working to scale the system in multiple dimensions. So one is we wanted to make our index bigger, so we could retrieve from a larger index, which always helps your quality in general. Uh, because if you don't have the page in your index, you're going to not do well. Um, and then we also needed to scale our capacity because we were, our traffic was growing quite extensively. Um, and so we had, you know, a sharded system where you have more and more shards as the index grows, you have like 30 shards. And then if you want to double the index size, you make 60 shards so that you can bound the latency by which you respond for any particular user query. Um, and then as traffic grows, you add, you add more and more replicas of each of those. And so we eventually did the math that realized that in a data center where we had say 60 shards and, um, you know, 20 copies of each shard, we now had 1200 machines, uh, with disks. And we did the math and we're like, Hey, one copy of that index would actually fit in memory across 1200 machines. So in 2001, we introduced, uh, we put our entire index in memory and what that enabled from a quality perspective was amazing. Um, and so we had more and more replicas of each of those. Before you had to be really careful about, you know, how many different terms you looked at for a query, because every one of them would involve a disk seek on every one of the 60 shards. And so you, as you make your index bigger, that becomes even more inefficient. But once you have the whole index in memory, it's totally fine to have 50 terms you throw into the query from the user's original three or four word query, because now you can add synonyms like restaurant and restaurants and cafe and, uh, you know, things like that. Uh, bistro and all these things. And you can suddenly start, uh, sort of really, uh, getting at the meaning of the word as opposed to the exact semantic form the user typed in. And that was, you know, 2001, very much pre LLM, but really it was about softening the, the strict definition of what the user typed in order to get at the meaning.Alessio Fanelli [00:26:47]: What are like principles that you use to like design the systems, especially when you have, I mean, in 2001, the internet is like. Doubling, tripling every year in size is not like, uh, you know, and I think today you kind of see that with LLMs too, where like every year the jumps in size and like capabilities are just so big. Are there just any, you know, principles that you use to like, think about this? Yeah.Jeff Dean [00:27:08]: I mean, I think, uh, you know, first, whenever you're designing a system, you want to understand what are the sort of design parameters that are going to be most important in designing that, you know? So, you know, how many queries per second do you need to handle? How big is the internet? How big is the index you need to handle? How much data do you need to keep for every document in the index? How are you going to look at it when you retrieve things? Um, what happens if traffic were to double or triple, you know, will that system work well? And I think a good design principle is you're going to want to design a system so that the most important characteristics could scale by like factors of five or 10, but probably not beyond that because often what happens is if you design a system for X. And something suddenly becomes a hundred X, that would enable a very different point in the design space that would not make sense at X. But all of a sudden at a hundred X makes total sense. So like going from a disk space index to a in memory index makes a lot of sense once you have enough traffic, because now you have enough replicas of the sort of state on disk that those machines now actually can hold, uh, you know, a full copy of the, uh, index and memory. Yeah. And that all of a sudden enabled. A completely different design that wouldn't have been practical before. Yeah. Um, so I'm, I'm a big fan of thinking through designs in your head, just kind of playing with the design space a little before you actually do a lot of writing of code. But, you know, as you said, in the early days of Google, we were growing the index, uh, quite extensively. We were growing the update rate of the index. So the update rate actually is the parameter that changed the most. Surprising. So it used to be once a month.Shawn Wang [00:28:55]: Yeah.Jeff Dean [00:28:56]: And then we went to a system that could update any particular page in like sub one minute. Okay.Shawn Wang [00:29:02]: Yeah. Because this is a competitive advantage, right?Jeff Dean [00:29:04]: Because all of a sudden news related queries, you know, if you're, if you've got last month's news index, it's not actually that useful for.Shawn Wang [00:29:11]: News is a special beast. Was there any, like you could have split it onto a separate system.Jeff Dean [00:29:15]: Well, we did. We launched a Google news product, but you also want news related queries that people type into the main index to also be sort of updated.Shawn Wang [00:29:23]: So, yeah, it's interesting. And then you have to like classify whether the page is, you have to decide which pages should be updated and what frequency. Oh yeah.Jeff Dean [00:29:30]: There's a whole like, uh, system behind the scenes that's trying to decide update rates and importance of the pages. So even if the update rate seems low, you might still want to recrawl important pages quite often because, uh, the likelihood they change might be low, but the value of having updated is high.Shawn Wang [00:29:50]: Yeah, yeah, yeah, yeah. Uh, well, you know, yeah. This, uh, you know, mention of latency and, and saving things to this reminds me of one of your classics, which I have to bring up, which is latency numbers. Every programmer should know, uh, was there a, was it just a, just a general story behind that? Did you like just write it down?Jeff Dean [00:30:06]: I mean, this has like sort of eight or 10 different kinds of metrics that are like, how long does a cache mistake? How long does branch mispredict take? How long does a reference domain memory take? How long does it take to send, you know, a packet from the U S to the Netherlands or something? Um,Shawn Wang [00:30:21]: why Netherlands, by the way, or is it, is that because of Chrome?Jeff Dean [00:30:25]: Uh, we had a data center in the Netherlands, um, so, I mean, I think this gets to the point of being able to do the back of the envelope calculations. So these are sort of the raw ingredients of those, and you can use them to say, okay, well, if I need to design a system to do image search and thumb nailing or something of the result page, you know, how, what I do that I could pre-compute the image thumbnails. I could like. Try to thumbnail them on the fly from the larger images. What would that do? How much dis bandwidth than I need? How many des seeks would I do? Um, and you can sort of actually do thought experiments in, you know, 30 seconds or a minute with the sort of, uh, basic, uh, basic numbers at your fingertips. Uh, and then as you sort of build software using higher level libraries, you kind of want to develop the same intuitions for how long does it take to, you know, look up something in this particular kind of.Shawn Wang [00:31:21]: I'll see you next time.Shawn Wang [00:31:51]: Which is a simple byte conversion. That's nothing interesting. I wonder if you have any, if you were to update your...Jeff Dean [00:31:58]: I mean, I think it's really good to think about calculations you're doing in a model, either for training or inference.Jeff Dean [00:32:09]: Often a good way to view that is how much state will you need to bring in from memory, either like on-chip SRAM or HBM from the accelerator. Attached memory or DRAM or over the network. And then how expensive is that data motion relative to the cost of, say, an actual multiply in the matrix multiply unit? And that cost is actually really, really low, right? Because it's order, depending on your precision, I think it's like sub one picodule.Shawn Wang [00:32:50]: Oh, okay. You measure it by energy. Yeah. Yeah.Jeff Dean [00:32:52]: Yeah. I mean, it's all going to be about energy and how do you make the most energy efficient system. And then moving data from the SRAM on the other side of the chip, not even off the off chip, but on the other side of the same chip can be, you know, a thousand picodules. Oh, yeah. And so all of a sudden, this is why your accelerators require batching. Because if you move, like, say, the parameter of a model from SRAM on the, on the chip into the multiplier unit, that's going to cost you a thousand picodules. So you better make use of that, that thing that you moved many, many times with. So that's where the batch dimension comes in. Because all of a sudden, you know, if you have a batch of 256 or something, that's not so bad. But if you have a batch of one, that's really not good.Shawn Wang [00:33:40]: Yeah. Yeah. Right.Jeff Dean [00:33:41]: Because then you paid a thousand picodules in order to do your one picodule multiply.Shawn Wang [00:33:46]: I have never heard an energy-based analysis of batching.Jeff Dean [00:33:50]: Yeah. I mean, that's why people batch. Yeah. Ideally, you'd like to use batch size one because the latency would be great.Shawn Wang [00:33:56]: The best latency.Jeff Dean [00:33:56]: But the energy cost and the compute cost inefficiency that you get is quite large. So, yeah.Shawn Wang [00:34:04]: Is there a similar trick like, like, like you did with, you know, putting everything in memory? Like, you know, I think obviously NVIDIA has caused a lot of waves with betting very hard on SRAM with Grok. I wonder if, like, that's something that you already saw with, with the TPUs, right? Like that, that you had to. Uh, to serve at your scale, uh, you probably sort of saw that coming. Like what, what, what hardware, uh, innovations or insights were formed because of what you're seeing there?Jeff Dean [00:34:33]: Yeah. I mean, I think, you know, TPUs have this nice, uh, sort of regular structure of 2D or 3D meshes with a bunch of chips connected. Yeah. And each one of those has HBM attached. Um, I think for serving some kinds of models, uh, you know, you, you pay a lot higher cost. Uh, and time latency, um, bringing things in from HBM than you do bringing them in from, uh, SRAM on the chip. So if you have a small enough model, you can actually do model parallelism, spread it out over lots of chips and you actually get quite good throughput improvements and latency improvements from doing that. And so you're now sort of striping your smallish scale model over say 16 or 64 chips. Uh, but as if you do that and it all fits in. In SRAM, uh, that can be a big win. So yeah, that's not a surprise, but it is a good technique.Alessio Fanelli [00:35:27]: Yeah. What about the TPU design? Like how much do you decide where the improvements have to go? So like, this is like a good example of like, is there a way to bring the thousand picojoules down to 50? Like, is it worth designing a new chip to do that? The extreme is like when people say, oh, you should burn the model on the ASIC and that's kind of like the most extreme thing. How much of it? Is it worth doing an hardware when things change so quickly? Like what was the internal discussion? Yeah.Jeff Dean [00:35:57]: I mean, we, we have a lot of interaction between say the TPU chip design architecture team and the sort of higher level modeling, uh, experts, because you really want to take advantage of being able to co-design what should future TPUs look like based on where we think the sort of ML research puck is going, uh, in some sense, because, uh, you know, as a hardware designer for ML and in particular, you're trying to design a chip starting today and that design might take two years before it even lands in a data center. And then it has to sort of be a reasonable lifetime of the chip to take you three, four or five years. So you're trying to predict two to six years out where, what ML computations will people want to run two to six years out in a very fast changing field. And so having people with interest. Interesting ML research ideas of things we think will start to work in that timeframe or will be more important in that timeframe, uh, really enables us to then get, you know, interesting hardware features put into, you know, TPU N plus two, where TPU N is what we have today.Shawn Wang [00:37:10]: Oh, the cycle time is plus two.Jeff Dean [00:37:12]: Roughly. Wow. Because, uh, I mean, sometimes you can squeeze some changes into N plus one, but, you know, bigger changes are going to require the chip. Yeah. Design be earlier in its lifetime design process. Um, so whenever we can do that, it's generally good. And sometimes you can put in speculative features that maybe won't cost you much chip area, but if it works out, it would make something, you know, 10 times as fast. And if it doesn't work out, well, you burned a little bit of tiny amount of your chip area on that thing, but it's not that big a deal. Uh, sometimes it's a very big change and we want to be pretty sure this is going to work out. So we'll do like lots of carefulness. Uh, ML experimentation to show us, uh, this is actually the, the way we want to go. Yeah.Alessio Fanelli [00:37:58]: Is there a reverse of like, we already committed to this chip design so we can not take the model architecture that way because it doesn't quite fit?Jeff Dean [00:38:06]: Yeah. I mean, you, you definitely have things where you're going to adapt what the model architecture looks like so that they're efficient on the chips that you're going to have for both training and inference of that, of that, uh, generation of model. So I think it kind of goes both ways. Um, you know, sometimes you can take advantage of, you know, lower precision things that are coming in a future generation. So you can, might train it at that lower precision, even if the current generation doesn't quite do that. Mm.Shawn Wang [00:38:40]: Yeah. How low can we go in precision?Jeff Dean [00:38:43]: Because people are saying like ternary is like, uh, yeah, I mean, I'm a big fan of very low precision because I think that gets, that saves you a tremendous amount of time. Right. Because it's picojoules per bit that you're transferring and reducing the number of bits is a really good way to, to reduce that. Um, you know, I think people have gotten a lot of luck, uh, mileage out of having very low bit precision things, but then having scaling factors that apply to a whole bunch of, uh, those, those weights. Scaling. How does it, how does it, okay.Shawn Wang [00:39:15]: Interesting. You, so low, low precision, but scaled up weights. Yeah. Huh. Yeah. Never considered that. Yeah. Interesting. Uh, w w while we're on this topic, you know, I think there's a lot of, um, uh, this, the concept of precision at all is weird when we're sampling, you know, uh, we just, at the end of this, we're going to have all these like chips that I'll do like very good math. And then we're just going to throw a random number generator at the start. So, I mean, there's a movement towards, uh, energy based, uh, models and processors. I'm just curious if you've, obviously you've thought about it, but like, what's your commentary?Jeff Dean [00:39:50]: Yeah. I mean, I think. There's a bunch of interesting trends though. Energy based models is one, you know, diffusion based models, which don't sort of sequentially decode tokens is another, um, you know, speculative decoding is a way that you can get sort of an equivalent, very small.Shawn Wang [00:40:06]: Draft.Jeff Dean [00:40:07]: Batch factor, uh, for like you predict eight tokens out and that enables you to sort of increase the effective batch size of what you're doing by a factor of eight, even, and then you maybe accept five or six of those tokens. So you get. A five, a five X improvement in the amortization of moving weights, uh, into the multipliers to do the prediction for the, the tokens. So these are all really good techniques and I think it's really good to look at them from the lens of, uh, energy, real energy, not energy based models, um, and, and also latency and throughput, right? If you look at things from that lens, that sort of guides you to. Two solutions that are gonna be, uh, you know, better from, uh, you know, being able to serve larger models or, you know, equivalent size models more cheaply and with lower latency.Shawn Wang [00:41:03]: Yeah. Well, I think, I think I, um, it's appealing intellectually, uh, haven't seen it like really hit the mainstream, but, um, I do think that, uh, there's some poetry in the sense that, uh, you know, we don't have to do, uh, a lot of shenanigans if like we fundamentally. Design it into the hardware. Yeah, yeah.Jeff Dean [00:41:23]: I mean, I think there's still a, there's also sort of the more exotic things like analog based, uh, uh, computing substrates as opposed to digital ones. Uh, I'm, you know, I think those are super interesting cause they can be potentially low power. Uh, but I think you often end up wanting to interface that with digital systems and you end up losing a lot of the power advantages in the digital to analog and analog to digital conversions. You end up doing, uh, at the sort of boundaries. And periphery of that system. Um, I still think there's a tremendous distance we can go from where we are today in terms of energy efficiency with sort of, uh, much better and specialized hardware for the models we care about.Shawn Wang [00:42:05]: Yeah.Alessio Fanelli [00:42:06]: Um, any other interesting research ideas that you've seen, or like maybe things that you cannot pursue a Google that you would be interested in seeing researchers take a step at, I guess you have a lot of researchers. Yeah, I guess you have enough, but our, our research.Jeff Dean [00:42:21]: Our research portfolio is pretty broad. I would say, um, I mean, I think, uh, in terms of research directions, there's a whole bunch of, uh, you know, open problems and how do you make these models reliable and able to do much longer, kind of, uh, more complex tasks that have lots of subtasks. How do you orchestrate, you know, maybe one model that's using other models as tools in order to sort of build, uh, things that can accomplish, uh, you know, much more. Yeah. Significant pieces of work, uh, collectively, then you would ask a single model to do. Um, so that's super interesting. How do you get more verifiable, uh, you know, how do you get RL to work for non-verifiable domains? I think it's a pretty interesting open problem because I think that would broaden out the capabilities of the models, the improvements that you're seeing in both math and coding. Uh, if we could apply those to other less verifiable domains, because we've come up with RL techniques that actually enable us to do that. Uh, effectively, that would, that would really make the models improve quite a lot. I think.Alessio Fanelli [00:43:26]: I'm curious, like when we had Noam Brown on the podcast, he said, um, they already proved you can do it with deep research. Um, you kind of have it with AI mode in a way it's not verifiable. I'm curious if there's any thread that you think is interesting there. Like what is it? Both are like information retrieval of JSON. So I wonder if it's like the retrieval is like the verifiable part. That you can score or what are like, yeah, yeah. How, how would you model that, that problem?Jeff Dean [00:43:55]: Yeah. I mean, I think there are ways of having other models that can evaluate the results of what a first model did, maybe even retrieving. Can you have another model that says, is this things, are these things you retrieved relevant? Or can you rate these 2000 things you retrieved to assess which ones are the 50 most relevant or something? Um, I think those kinds of techniques are actually quite effective. Sometimes I can even be the same model, just prompted differently to be a, you know, a critic as opposed to a, uh, actual retrieval system. Yeah.Shawn Wang [00:44:28]: Um, I do think like there, there is that, that weird cliff where like, it feels like we've done the easy stuff and then now it's, but it always feels like that every year. It's like, oh, like we know, we know, and the next part is super hard and nobody's figured it out. And, uh, exactly with this RLVR thing where like everyone's talking about, well, okay, how do we. the next stage of the non-verifiable stuff. And everyone's like, I don't know, you know, Ellen judge.Jeff Dean [00:44:56]: I mean, I feel like the nice thing about this field is there's lots and lots of smart people thinking about creative solutions to some of the problems that we all see. Uh, because I think everyone sort of sees that the models, you know, are great at some things and they fall down around the edges of those things and, and are not as capable as we'd like in those areas. And then coming up with good techniques and trying those. And seeing which ones actually make a difference is sort of what the whole research aspect of this field is, is pushing forward. And I think that's why it's super interesting. You know, if you think about two years ago, we were struggling with GSM, eight K problems, right? Like, you know, Fred has two rabbits. He gets three more rabbits. How many rabbits does he have? That's a pretty far cry from the kinds of mathematics that the models can, and now you're doing IMO and Erdos problems in pure language. Yeah. Yeah. Pure language. So that is a really, really amazing jump in capabilities in, you know, in a year and a half or something. And I think, um, for other areas, it'd be great if we could make that kind of leap. Uh, and you know, we don't exactly see how to do it for some, some areas, but we do see it for some other areas and we're going to work hard on making that better. Yeah.Shawn Wang [00:46:13]: Yeah.Alessio Fanelli [00:46:14]: Like YouTube thumbnail generation. That would be very helpful. We need that. That would be AGI. We need that.Shawn Wang [00:46:20]: That would be. As far as content creators go.Jeff Dean [00:46:22]: I guess I'm not a YouTube creator, so I don't care that much about that problem, but I guess, uh, many people do.Shawn Wang [00:46:27]: It does. Yeah. It doesn't, it doesn't matter. People do judge books by their covers as it turns out. Um, uh, just to draw a bit on the IMO goal. Um, I'm still not over the fact that a year ago we had alpha proof and alpha geometry and all those things. And then this year we were like, screw that we'll just chuck it into Gemini. Yeah. What's your reflection? Like, I think this, this question about. Like the merger of like symbolic systems and like, and, and LMS, uh, was a very much core belief. And then somewhere along the line, people would just said, Nope, we'll just all do it in the LLM.Jeff Dean [00:47:02]: Yeah. I mean, I think it makes a lot of sense to me because, you know, humans manipulate symbols, but we probably don't have like a symbolic representation in our heads. Right. We have some distributed representation that is neural net, like in some way of lots of different neurons. And activation patterns firing when we see certain things and that enables us to reason and plan and, you know, do chains of thought and, you know, roll them back now that, that approach for solving the problem doesn't seem like it's going to work. I'm going to try this one. And, you know, in a lot of ways we're emulating what we intuitively think, uh, is happening inside real brains in neural net based models. So it never made sense to me to have like completely separate. Uh, discrete, uh, symbolic things, and then a completely different way of, of, uh, you know, thinking about those things.Shawn Wang [00:47:59]: Interesting. Yeah. Uh, I mean, it's maybe seems obvious to you, but it wasn't obvious to me a year ago. Yeah.Jeff Dean [00:48:06]: I mean, I do think like that IMO with, you know, translating to lean and using lean and then the next year and also a specialized geometry model. And then this year switching to a single unified model. That is roughly the production model with a little bit more inference budget, uh, is actually, you know, quite good because it shows you that the capabilities of that general model have improved dramatically and, and now you don't need the specialized model. This is actually sort of very similar to the 2013 to 16 era of machine learning, right? Like it used to be, people would train separate models for lots of different, each different problem, right? I have, I want to recognize street signs and something. So I train a street sign. Recognition recognition model, or I want to, you know, decode speech recognition. I have a speech model, right? I think now the era of unified models that do everything is really upon us. And the question is how well do those models generalize to new things they've never been asked to do and they're getting better and better.Shawn Wang [00:49:10]: And you don't need domain experts. Like one of my, uh, so I interviewed ETA who was on, who was on that team. Uh, and he was like, yeah, I, I don't know how they work. I don't know where the IMO competition was held. I don't know the rules of it. I just trained the models, the training models. Yeah. Yeah. And it's kind of interesting that like people with these, this like universal skill set of just like machine learning, you just give them data and give them enough compute and they can kind of tackle any task, which is the bitter lesson, I guess. I don't know. Yeah.Jeff Dean [00:49:39]: I mean, I think, uh, general models, uh, will win out over specialized ones in most cases.Shawn Wang [00:49:45]: Uh, so I want to push there a bit. I think there's one hole here, which is like, uh. There's this concept of like, uh, maybe capacity of a model, like abstractly a model can only contain the number of bits that it has. And, uh, and so it, you know, God knows like Gemini pro is like one to 10 trillion parameters. We don't know, but, uh, the Gemma models, for example, right? Like a lot of people want like the open source local models that are like that, that, that, and, and, uh, they have some knowledge, which is not necessary, right? Like they can't know everything like, like you have the. The luxury of you have the big model and big model should be able to capable of everything. But like when, when you're distilling and you're going down to the small models, you know, you're actually memorizing things that are not useful. Yeah. And so like, how do we, I guess, do we want to extract that? Can we, can we divorce knowledge from reasoning, you know?Jeff Dean [00:50:38]: Yeah. I mean, I think you do want the model to be most effective at reasoning if it can retrieve things, right? Because having the model devote precious parameter space. To remembering obscure facts that could be looked up is actually not the best use of that parameter space, right? Like you might prefer something that is more generally useful in more settings than this obscure fact that it has. Um, so I think that's always attention at the same time. You also don't want your model to be kind of completely detached from, you know, knowing stuff about the world, right? Like it's probably useful to know how long the golden gate be. Bridges just as a general sense of like how long are bridges, right? And, uh, it should have that kind of knowledge. It maybe doesn't need to know how long some teeny little bridge in some other more obscure part of the world is, but, uh, it does help it to have a fair bit of world knowledge and the bigger your model is, the more you can have. Uh, but I do think combining retrieval with sort of reasoning and making the model really good at doing multiple stages of retrieval. Yeah.Shawn Wang [00:51:49]: And reasoning through the intermediate retrieval results is going to be a, a pretty effective way of making the model seem much more capable, because if you think about, say, a personal Gemini, yeah, right?Jeff Dean [00:52:01]: Like we're not going to train Gemini on my email. Probably we'd rather have a single model that, uh, we can then use and use being able to retrieve from my email as a tool and have the model reason about it and retrieve from my photos or whatever, uh, and then make use of that and have multiple. Um, you know, uh, stages of interaction. that makes sense.Alessio Fanelli [00:52:24]: Do you think the vertical models are like, uh, interesting pursuit? Like when people are like, oh, we're building the best healthcare LLM, we're building the best law LLM, are those kind of like short-term stopgaps or?Jeff Dean [00:52:37]: No, I mean, I think, I think vertical models are interesting. Like you want them to start from a pretty good base model, but then you can sort of, uh, sort of viewing them, view them as enriching the data. Data distribution for that particular vertical domain for healthcare, say, um, we're probably not going to train or for say robotics. We're probably not going to train Gemini on all possible robotics data. We, you could train it on because we want it to have a balanced set of capabilities. Um, so we'll expose it to some robotics data, but if you're trying to build a really, really good robotics model, you're going to want to start with that and then train it on more robotics data. And then maybe that would. It's multilingual translation capability, but improve its robotics capabilities. And we're always making these kind of, uh, you know, trade-offs in the data mix that we train the base Gemini models on. You know, we'd love to include data from 200 more languages and as much data as we have for those languages, but that's going to displace some other capabilities of the model. It won't be as good at, um, you know, Pearl programming, you know, it'll still be good at Python programming. Cause we'll include it. Enough. Of that, but there's other long tail computer languages or coding capabilities that it may suffer on or multi, uh, multimodal reasoning capabilities may suffer. Cause we didn't get to expose it to as much data there, but it's really good at multilingual things. So I, I think some combination of specialized models, maybe more modular models. So it'd be nice to have the capability to have those 200 languages, plus this awesome robotics model, plus this awesome healthcare, uh, module that all can be knitted together to work in concert and called upon in different circumstances. Right? Like if I have a health related thing, then it should enable using this health module in conjunction with the main base model to be even better at those kinds of things. Yeah.Shawn Wang [00:54:36]: Installable knowledge. Yeah.Jeff Dean [00:54:37]: Right.Shawn Wang [00:54:38]: Just download as a, as a package.Jeff Dean [00:54:39]: And some of that installable stuff can come from retrieval, but some of it probably should come from preloaded training on, you know, uh, a hundred billion tokens or a trillion tokens of health data. Yeah.Shawn Wang [00:54:51]: And for listeners, I think, uh, I will highlight the Gemma three end paper where they, there was a little bit of that, I think. Yeah.Alessio Fanelli [00:54:56]: Yeah. I guess the question is like, how many billions of tokens do you need to outpace the frontier model improvements? You know, it's like, if I have to make this model better healthcare and the main. Gemini model is still improving. Do I need 50 billion tokens? Can I do it with a hundred, if I need a trillion healthcare tokens, it's like, they're probably not out there that you don't have, you know, I think that's really like the.Jeff Dean [00:55:21]: Well, I mean, I think healthcare is a particularly challenging domain, so there's a lot of healthcare data that, you know, we don't have access to appropriately, but there's a lot of, you know, uh, healthcare organizations that want to train models on their own data. That is not public healthcare data, uh, not public health. But public healthcare data. Um, so I think there are opportunities there to say, partner with a large healthcare organization and train models for their use that are going to be, you know, more bespoke, but probably, uh, might be better than a general model trained on say, public data. Yeah.Shawn Wang [00:55:58]: Yeah. I, I believe, uh, by the way, also this is like somewhat related to the language conversation. Uh, I think one of your, your favorite examples was you can put a low resource language in the context and it just learns. Yeah.Jeff Dean [00:56:09]: Oh, yeah, I think the example we used was Calamon, which is truly low resource because it's only spoken by, I think 120 people in the world and there's no written text.Shawn Wang [00:56:20]: So, yeah. So you can just do it that way. Just put it in the context. Yeah. Yeah. But I think your whole data set in the context, right.Jeff Dean [00:56:27]: If you, if you take a language like, uh, you know, Somali or something, there is a fair bit of Somali text in the world that, uh, or Ethiopian Amharic or something, um, you know, we probably. Yeah. Are not putting all the data from those languages into the Gemini based training. We put some of it, but if you put more of it, you'll improve the capabilities of those models.Shawn Wang [00:56:49]: Yeah.Jeff Dean [00:56:49]:

Proton Drive Angebot

Graphics processing units (GPUs) have become the most important commodity in the AI boom — and have made Nvidia a multi-trillion dollar company. But the tensor processing unit (TPU) could present itself as competition for the GPU.TPUs are developed by Google specifically for AI workloads. And so far, Anthropic, OpenAI and Meta have reportedly made deals for Google's TPUs.Christopher Miller, historian at Tufts University and author of "Chip War: The Fight for the World's Most Critical Technology," explains what this could mean.

Graphics processing units (GPUs) have become the most important commodity in the AI boom — and have made Nvidia a multi-trillion dollar company. But the tensor processing unit (TPU) could present itself as competition for the GPU.TPUs are developed by Google specifically for AI workloads. And so far, Anthropic, OpenAI and Meta have reportedly made deals for Google's TPUs.Christopher Miller, historian at Tufts University and author of "Chip War: The Fight for the World's Most Critical Technology," explains what this could mean.

【PR：Genspark】 #Genspark #WorkwithGenspark「Claude Opus 4.6」も対象に！生成AIエージェント「Genspark」が使えるリンクはこちら！https://www.genspark.ai/?utm_source=yt&utm_campaign=yurucom※2026年内は、AIチャットおよびAI画像生成機能が無制限で利用可能。◯グローバル公式アカウント・X → https://x.com/genspark_ai・YouTube → https://www.youtube.com/@GensparkProduct◯日本公式アカウント・X → https://x.com/genspark_japan・YouTube → https://www.youtube.com/@Genspark-JapanGoogleが開発したTPUは異端すぎた。現代コンピュータの常識から逸脱した計算機の正体に迫りました。【目次】0:00 Googleからの刺客「TPU」5:07 TPUの強み「記憶しない」13:32 常識から逸脱した計算機20:10 現代コンピュータ、汎用から専用へ25:19 もっとなんでもやらせたい！28:24 結局、何がどう専用なの？31:19 TPUは生物模倣？34:56 Google株は買い？36:03 スライド最強AIエージェント Genspark【参考文献】◯ご冗談でしょう、ファインマンさん（上）https://amzn.to/49TrZM7◯AIチップ開発競争、グーグルが猛追－「エヌビディア1強」に風穴も（Bloomberg）https://www.bloomberg.com/jp/news/articles/2025-11-25/T69EB2KJH6VD00◯Computing's Energy Problem: (and what we can do about it)https://pdfs.semanticscholar.org/9476/20a1854655ed91a86b90d12695e05be85983.pdf◯Gemini 3 Pro – Model Cardhttps://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-Pro-Model-Card.pdf◯In-Datacenter Performance Analysis of a Tensor Processing Unithttps://arxiv.org/pdf/1704.04760◯An in-depth look at Google's first Tensor Processing Unit (TPU)https://cloud.google.com/blog/products/ai-machine-learning/an-in-depth-look-at-googles-first-tensor-processing-unit-tpu◯Matrix Multiplication Background User's Guidehttps://docs.nvidia.com/deeplearning/performance/dl-performance-matrix-multiplication/index.html◯A New Golden Age for Computer Architecturehttps://www.doc.ic.ac.uk/~wl/teachlocal/arch/papers/cacm19golden-age.pdf【サポーターコミュニティへの加入はこちらから！】⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://yurugengo.com/support⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠【親チャンネル：ゆる言語学ラジオ】⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.youtube.com/@yurugengo⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠【実店舗プロジェクト：ゆる学徒カフェ】⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.youtube.com/@yurugakuto⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠【おたよりフォーム】⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://forms.gle/BLEZpLcdEPmoZTH4A⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠※皆様からの楽しいおたよりをお待ちしています！【お仕事依頼はこちら！】info@pedantic.jp【堀元見プロフィール】慶應義塾大学理工学部卒。専攻は情報工学。理屈っぽいコンテンツを作り散らかすことで生計を立てている。Twitter→⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://twitter.com/kenhori2⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠noteマガジン→⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://note.com/kenhori2/m/m125fc4524aca⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠個人YouTube→⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.youtube.com/@kenHorimoto⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠【水野太貴プロフィール】1995年生まれ。愛知県出身。名古屋大学文学部卒。専攻は言語学。本業は雑誌編集者。著書に『会話の0.2秒を言語学する 』（新潮社）などがある。Podcast「神保町で会いましょう」のパーソナリティも務める。Twitter→⁠⁠https://x.com/yuru_mizuno⁠⁠神保町で会いましょう→⁠⁠https://open.spotify.com/show/6cYkvDO0HnJKLPgDBGUjjS

C'est une annonce fracassante, et elle nous vient de Microsoft.Le géant de Redmond vient de dévoiler la puce Maia 200, un processeur maison capable de faire trembler le monopole de Nvidia sur le marché de l'intelligence artificielle.Une puce déjà en productionCette nouvelle puce ne se contente pas de promesses techniques sur le papier. Elle est déjà opérationnelle et propulse actuellement les tâches d'inférence réalisées avec GPT-5.2 en production dans les datacenters de Microsoft.Scott Guthrie, le big boss de l'IA et du cloud chez Microsoft, assure aussi que Maia 200 sera utilisé en interne pour la génération de données synthétiques pour permettre avec de l'apprentissage par renforcement d'améliorer les nouveaux modèles d'IA.C'est un signal fort envoyé au marché, puisque Nvidia contrôle aujourd'hui environ 95 % du secteur des puces IA avec des marges dépassant les 70 %. Pour les professionnels du cloud et du développement, l'arrivée d'une alternative comme celle de Microsoft signifie potentiellement une réduction de la dépendance vis-à-vis d'un fournisseur unique et donc une optimisation des coûts d'infrastructure.Collaboration avec TSMCTechniquement, la puce Maia 200 affiche des caractéristiques impressionnantes. Gravée en 3 nanomètres en collaboration avec TSMC, elle embarque 140 milliards de transistors et délivre une puissance de 10 pétaflops en précision FP4.Côté mémoire, elle dispose de 216 gigaoctets de mémoire à haute bande passante, permettant de traiter des flux de données à une vitesse de 7 téraoctets par seconde.Ces chiffres placent la Maia 200 comme un concurrent direct des meilleures solutions actuelles, offrant une capacité de calcul massive pour les modèles de langage d'IA les plus gourmands.En concurrence avec Trainium 3 et TPU version 7 Enfin, l'enjeu majeur pour Microsoft est de prouver l'efficacité réelle de son matériel face à la concurrence. Les premiers tests indiquent que la Maia 200 serait trois fois plus rapide que la puce Trainium 3 d'Amazon et surpasserait le TPU version 7 de Google sur certains calculs.Plus important encore pour les budgets IT, Microsoft annonce un rapport performance-prix supérieur de 30 % par rapport à la génération actuelle.Pour accompagner ce lancement, une version préliminaire du kit de développement est déjà accessible aux développeurs et aux chercheurs.Le ZD Tech est sur toutes les plateformes de podcast ! Abonnez-vous !Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.

Full Audio at https://podcasts.apple.com/us/podcast/ai-business-and-development-daily-news-rundown/id1684415169?i=1000746957648

Murderbot is ADG-nominated for Episode 1: “FreeCommerce” — and the production design and set decoration work is a huge part of what makes the show feel so specific, lived-in, and visually smart.Discover the sci-fi production design magic behind the ADG-Nominated Murderbot with Production Designer Sue Chanand Set Decorator Rob Hepburn. From inflatable TPU solar habitats to 3D-printed med labs, they reveal how they brought a warm, lived-in aesthetic to space. Hosted by Kim Wannop, this episode dives into sourcing, greenscaping, LED tech, and the wild storytelling choices that made Murderbot unforgettable.

There are more running shoe brands and models than ever - and all the jargon can feel confusing. In this episode, we help you navigate what features of a running shoe actually matter - and what's just good marketing. You'll learn about supershoes vs supertrainers, why heel to toe drop matters, what the different foams do, and more.Thank you to our sponsors:✨ Amazfit: User-friendly simple running watches with advanced features, at an affordable price point. Use link http://bit.ly/4nai73H for 10% off your purchase.✨Title Nine: Comfortable sports bras that actually fit, from a women-owned company. Use code RUNTOTHEFINISH for free shipping at https://runtothefinish.com/title-nine/✨Join us on Patreon.com/treadlightlyrunning or subscribe on Apple Podcasts for special subscriber-only content!In this episode, you'll learn:✅ The difference between carbon and nylon plated shoes✅ Why you shouldn't train in supershoes all the time✅ How to safely introduce carbon plated shoes✅ The pros and cons of high stack height shoes✅ The most important features to consider when buying new running shoes✅ Understanding PEBA, TPU, and EVA foams✅ Do you need a running shoe rotation?If you enjoyed this episode, you may also like:

Send us a textWe trace how affordable, reliable material extrusion is changing prosthetics and orthotics—from student labs to jungle clinics—and why toolpaths, not just materials, will drive the next gains in comfort, strength and cost. Real patient stories show the economics and ethics of access at scale.• season launch and mission to improve patient outcomes• shift from tinkering to reliable, prosumer 3D printers• material extrusion vs FDM and why terminology matters• nonplanar layers, multimaterial potential, pellet economics• toward truly digital extrusion with better sensing and AI• application focus over generalization in O&P innovation• case study on low-cost pediatric prosthesis with reuse of CAD data• orthoses workflows moving toward “toaster-like” simplicity• education pathways as students learn on clinic-grade printers• materials outlook: TPU, TPE, silicone prospects, polycarbonate tradeoffs• variable density, air pockets, and hybrid fill strategies for comfort• polar kinematics and toolpath planning as the next frontier• print farms, software orchestration, and scaling productionSpecial thanks to Advanced 3D for sponsoring this episode.Support the show

In this episode, we are joined by Alessio Pagliai and Stepan Drunks. Brought to you by Sovol (https://jle.vi/sovol) and OctoEverywhere (https://octoeverywhere.com/welcome?id=podcast).

In this episode of the Phoenix Cast, hosts John and Kyle kick off 2026 with a jam-packed current events roundup covering the React to Shell vulnerability (think Log4Shell but for the front end), the Marine Corps' new drone training requirements, Google's TPU announcements that might have NVIDIA sweating, and the launch of GenAI.mil. They also share some exciting podcast milestones, dish out their 2026 predictions, and Kyle reveals his holiday vendetta against PowerPoint that resulted in building his own AI-powered presentation tool.We'd love to hear your thoughts! Tweet us at our new handle, @ThePhoenixCast, and don't forget to join our LinkedIn Group to connect with fellow Phoenix Casters. If you enjoyed the episode, help us out by leaving one of those coveted 5-star reviews on Apple Podcasts. Thanks for listening!LinksKyle's “The 8 Levels of AI Learning for Modern Commanders”https://www.linkedin.com/pulse/8-levels-ai-learning-modern-commanders-kyle-kmo-moschetto-mxuycReactShell:https://securityboulevard.com/2026/01/top-cves-of-december-2025/TorchTPU:https://hyperframeresearch.com/2025/12/24/can-googles-torchtpu-eventually-bridge-nvidias-cuda-moat/ WSJ: “Why AI Will Widen the Gap Between Superstars and Everybody Else”https://www.wsj.com/lifestyle/workplace/ai-workplace-tensions-what-to-do-c45f6b51?reflink=desktopwebshare_permalink USMC drone program:  https://www.marines.mil/News/Messages/Messages-Display/Article/4366306/approved-training-requirements-for-small-unmanned-aerial-systems/USMC AI WORKSHOP MARADMINhttps://www.marines.mil/News/Messages/Messages-Display/Article/4367572/united-states-marines-corps-generative-and-agentic-artificial-intelligence-work/II MEF Leadership AI:https://www.iimef.marines.mil/News/article-display/Article/4364616/ii-mef-advanced-ai-command-course/ Self-Paced AI Training (Military discount available)https://ftcg.io/self-paced-training Vibe Coding book (Gene Kim and Steve Yegge):https://itrevolution.com/product/vibe-coding-book/Gas Town:https://steve-yegge.medium.com/welcome-to-gas-town-4f25ee16dd04

Welcome to your weekly UAS News Update. We have three stories for you this week; leaked pricing for the upcoming DJI Avata 360, the world record for the fastest drone has been shattered, and public safety is starting the year off with a ton of drones for good stories! Let's get to it.First up thanks to a leaked pricing table from a Chinese retail store, we have what appears to be the final pricing for the DJI Avata 360. And yes, the Avata 360 is already FCC approved. Now for the prices. In China, the base drone is listed at ¥2,988, which is about $426 USD. The Standard Combo with the Motion Controller 3 is about $569, and the Fly More Combo comes in at around $811. That puts the estimated US price for the base drone around $489, and the Fly More Combo will likely land right at that classic DJI price point of $999.This drone is rumored to brings true spherical 360 capture to an FPV platform, which is a huge deal. There are also rumors it could be under 250 grams. It seems to be a direct challenger to the Insta360 Antigravity A1, and DJI is betting that immersive 360 FPV is compelling enough for people to swallow that price tag for this new tech. Next up, for all you speed demons and FPV builders out there, the record for the world's fastest drone has been absolutely demolished. Luke Maximo Bell and his team have reclaimed the Guinness World Record with their Peregreen V4 drone, clocking an official top speed of 408.60 miles per hour, or 657.59 kilometers per hour. They took the record back from Benjamin Biggs, who had set it at 389 mph.What's really impressive here is the engineering. They meticulously tested three different motors—the AOS Supernova 3220, the AMX 2826, and the T-Motor 3120. They ended up choosing the T-Motor 3120 not because it had the most thrust, but because it was the most reliable and ran cooler. That shows it's not just about peak power, but about surviving the run! The frame itself was 3D printed, merging a hard PETG material with a softer TPU on the nose cone. To get that extra speed, they also bumped the motor KV up from 800 to 900. I want to pause for just a minute to discuss an upcoming webinar we are hosting. This webinar is all about how to land clients in 7 days, and it's on Tuesday, January 13th. If you're struggling to get your first client, this is perfect for you. Be sure to preregister if you want to attend. Check out the link in the comments, and we'll see you there! Last up this week, we have a bunch of drones for good stories, out of several places across the country:- A hiker was rescued using a drone in Chillicothe, Ohio using a drone, likely using a DJI Matrice series.- A hiker in Oregon was rescued, likely using an M30T or Matrice 4T.- A man with dementia in Campbellsville KY was located using a Matrice 30T. - A Skydio X10 was used to capture a man in Wichita after an armed robbery.k- A DJI Matrice 400 was instrumental in a rescue in Michigan, after a snowmobile broke through lake ice, sending the two riders into the water. - And a Matrice 4T in Fishers, Indiana located a firearm after it was dumped by a suspect during a chase.These stories are proof that drones have become like any other tool for Public safety departments, and that they do save lives. Alright, that's it for this week, Join us in the premium community for Post flight, our uncensored show where we share our opinions, which aren't always suitable for YouTube! See you on Monday for the live! https://dronexl.co/2026/01/04/luke-bells-peregreen-v4-new-fastest-drone/https://dronexl.co/2026/01/02/dji-avata-360-price-china-us/https://dronexl.co/2025/12/31/police-drone-missing-hiker-ohio-search/https://dronexl.co/2026/01/05/dji-drone-ice-rescue-saginaw-bay/https://dronexl.co/2026/01/04/wichita-police-drone-robbery/https://dronexl.co/2026/01/04/dji-matrice-drone-campbellsville-missing/https://dronexl.co/2026/01/03/drone-rescue-lost-hiker-oregon/

In this episode of the Marathon Running Podcast, we sit down with Luca Ciccone, Director of Product Engineering at Saucony, to demystify the complex world of running shoe technology.Ever feel overwhelmed by the alphabet soup of shoe foams? We ask Luca to break down the science behind PEBA, EVA, TPU, and TPEE. He helps us understand which materials offer the most energy return, which are the most durable, and why certain shoes carry a higher price tag. Whether you are a casual jogger or a competitive marathoner, this deep dive into the "foam pyramid" provides the expert insights you need to make an informed decision on your next pair of trainers. Stay tuned for updates and expert insights to keep you informed on the latest in running and competitive sports.If you've ever wondered if "super foams" are worth the investment or why your shoes feel different after 200 miles, this episode provides the technical clarity you've been looking for from an industry veteran with over 20 years of experience.

בפרק היומי של "מפת החום – מהדורת אחד ביום", אני עושה סדר בכל מה שקרה ב־24 השעות האחרונות בעולם הכלכלה – מהשוק המקומי ועד הגלובלי.נושאים :1. סוגי השבבים שחשוב להכיר - CPU, GPU, TPU, ASIC2. ⁠הכנס של אינבידה שבוע שעברנתונים מאקרו־כלכליים, דיווחים חשובים, כותרות שזעזעו את השוק, דוחות כספיים של חברות, צעדים רגולטוריים, שינויים במדיניות ותחזיות מפתיעות – כל מה שצריך כדי להבין את התמונה המלאה.זהו פודקאסט קצר, חד ותמציתי – בלי רעש מיותר, רק תובנות פרקטיות וסקירה מקצועית של היום שהיה.הפרק מתעדכן מדי בוקר – ומעניק לכם יתרון אמיתי על שאר המשקיעים.לפתיחת חשבון מסחר במיטב:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://landing.meitav.co.il/he-IL/landing/trade/tradeleads?utm_source=%D7%92%D7%99%D7%90+%D7%A0%D7%AA%D7%9F&utm_medium=%D7%92%D7%99%D7%90+%D7%A0%D7%AA%D7%9F⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠לאינסטגרם שלי:⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.instagram.com/guynatan9/לאתר שלי:https://www.guynatan.com/

The lesson Austin Lyons says he will take away from 2025 is the deep ties AI has to the CapEx story. Companies like Nvidia (NVDA) and TSMC (TSM) are juggernauts he expects to benefit as AI spending accelerates in 2026. When it comes to growing TPU competition from Alphabet (GOOGL), he doesn't see it serving as a perfect substitute to Nvidia's GPUs. Austin later taps Apple's (AAPL) AI story, or lack thereof, and how it serves as a company bull case. ======== Schwab Network ========Empowering every investor and trader, every market day.Options involve risks and are not suitable for all investors. Before trading, read the Options Disclosure Document. http://bit.ly/2v9tH6DSubscribe to the Market Minute newsletter - https://schwabnetwork.com/subscribeDownload the iOS app - https://apps.apple.com/us/app/schwab-network/id1460719185Download the Amazon Fire Tv App - https://www.amazon.com/TD-Ameritrade-Network/dp/B07KRD76C7Watch on Sling - https://watch.sling.com/1/asset/191928615bd8d47686f94682aefaa007/watchWatch on Vizio - https://www.vizio.com/en/watchfreeplus-exploreWatch on DistroTV - https://www.distro.tv/live/schwab-network/Follow us on X – https://twitter.com/schwabnetworkFollow us on Facebook – https://www.facebook.com/schwabnetworkFollow us on LinkedIn - https://www.linkedin.com/company/schwab-network/About Schwab Network - https://schwabnetwork.com/about

In this final part of our series meeting with brands to highlight their updates in the next year, Andrea is joined by Russ Stevens & Carolyn Todd from Topo Athletic. They highlight the big redesign to the Specter 3 featuring an all new A-TPU midsole. They'r also bringing updates to the Atmos, Ultrafly, trail favorites like the Terraventure, and many more. Get your DOR Merch: https://doctors-of-running.myspreadshop.com/Get 20% off your first order from Skratch with code: DOCTORSOFRUNNING! https://www.skratchlabs.comChapters0:00 - Intro2:28 - What makes Topo tick6:08 - Specter 310:47 - Atmos 215:34 - Ultrafly 619:12 - Aura 222:48 - Connect28:28 - Ultraventure 531:18 - Terraventure 533:56 - Future Cyclone updates38:45 - What the next couple years have in store for Topo43:54 - Wrap-up

Aaron and Brian make some bold predictions for the 2026 Cloud and AI markets, as well as reviewing the biggest issues going into 2026.  SHOW: 989SHOW TRANSCRIPT: The Cloudcast #989 TranscriptSHOW VIDEO: https://youtube.com/@TheCloudcastNET CLOUD NEWS OF THE WEEK: http://bit.ly/cloudcast-cnotwCHECK OUT OUR NEW PODCAST: "CLOUDCAST BASICS"SHOW NOTES:CLOUD & AI NEWS OF THE MONTH - NOV 2025 (show)CLOUD & AI NEWS OF THE MONTH - OCT 2025 (show)CLOUD & AI NEWS OF THE MONTH - SEPT 2025 (show)CLOUD & AI NEWS OF THE MONTH - AUG 2025 (show)CLOUD & AI NEWS OF THE MONTH - JUL 2025 (show)CLOUD & AI NEWS OF THE MONTH - JUN 2025 (show)CLOUD & AI NEWS OF THE MONTH - MAY 2025 (show)CLOUD & AI NEWS OF THE MONTH - APR 2025 (show)CLOUD & AI NEWS OF THE MONTH - MAR 2025 (show)2026 CLOUD + AI PREDICTIONS (AND BIG ISSUES TO REVIEW)OpenAI Revenues and Focus AreasNVIDIA customer profitabilityCompanies moving to GOOG TPUsEnterprise success beyond CoPilot/GeminiEnterprise data+model trainabilityEnterprise price hikesBroadcom, AMD, Groq - alternative HW optionsData Center buildoutsDoes AI spending shiftWhat is Agentic AI?Long term spending + short term refocusesPREDICTIONS:At least one big AI IPO in 2026, and it won't go well. (Aaron says Anthropic)People will question whether Sam Altman is the right person to lead OpenAIAI will be a central issue in the 2026 US elections, either about job losses or electricity pricesOne major LPU/TPU/dedicated inference chip will break through in 2026Azure will be the Number One Cloud… (Aaron has to keep it going)We will start to see a shift in the Enterprise from big models in the sky (1+trillion parameters) to dedicated, purpose-built models of 500M or less in size for efficiency and securityGemini will dominate the consumer/prosumer space, OpenAI will go through the trough of disillusionmentThe industry will shift to a base/instruct and a reasoning split of modelsAWS and Azure will double down on being a solutions provider instead of a primitive supplier for AI and infrastructureFEEDBACK?Email: show at the cloudcast dot netTwitter/X: @cloudcastpodBlueSky: @cloudcastpod.bsky.socialInstagram: @cloudcastpodTikTok: @cloudcastpod

Turns out when the lights go out, Waymo's don't handle that well. Larry Ellison actually puts his money on the line. Somebody is pirating music like it's 1999. And two deep-dive looks at whether or not Google's TPU's really are a threat to Nvidia and OpenAI. Waymo resumes robotaxi service in San Francisco after blackout chaos — Musk says Tesla car service unaffected (CNBC) Paramount guarantees Larry Ellison backing in amended WBD bid (CNBC) Instacart Scraps All Price Tests After Customer Pushback (WSJ) Spotify Music Library Scraped by Pirate Activist Group (Billboard) ChatGPT will now let you pick how nice it is (The Verge) TPU Mania (The Chip Letter) Why Nvidia maintains its moat and Gemini won't kill OpenAI (SiliconANGLE) Learn more about your ad choices. Visit megaphone.fm/adchoices

Here it is! We review the year where scaling large AI models hit its ceiling, Google reclaimed momentum with efficient vertical integration, and the market shifted from hype to viability. Join us as we talk about why human-in-the-loop is failing, why generative AI agents validating other agents compounds errors, and how small expert data quietly beat the big models.• Google's resurgence with Gemini 3.0 and TPU-driven efficiency• Monetization pressures and ads in co-pilot assistants• Diminishing returns from LLM scaling• Human-in-the-loop pitfalls and incentives• Agents vs validation and compounding error• Small, high-quality data outperforming synthetic• Expert systems, causality, and interpretability• Research trends return toward statistical rigor• 2026 outlook for ROI, governance, and trustWe remain focused on the responsible use of AI. And while the market continues to adjust expectations for return on investment from AI, we're excited to see companies exploring "return on purpose" as the new foray into transformative AI systems for their business. What are you excited about for AI in 2026? What did you think? Let us know.Do you have a question or a discussion topic for the AI Fundamentalists? Connect with them to comment on your favorite topics: LinkedIn - Episode summaries, shares of cited articles, and more. YouTube - Was it something that we said? Good. Share your favorite quotes. Visit our page - see past episodes and submit your feedback! It continues to inspire future episodes.

Hour 4 opens with a recap of Senator Josh Hawley's earlier interview, touching on the president's speech, border security, and Amendment 3. The hosts discuss rumors of the Chicago Bears relocating and St. Louis stadium politics, then move to national headlines, including Archbishop Ronald Hicks' appointment in New York and Candace Owens' comments on a TPU scandal. The segment highlights DEI and LGBTQ training being forced on Illinois State Police, sparking controversy. The next segment features Shannon Bream analyzing the president's speech, economic updates, and international policy, including potential Venezuela strikes. Griff Jenkins takes the spotlight in the third segment, sharing personal holiday stories, travel plans, and insights on New York politics and progressive candidates. The final segment covers Democratic criticism of ICE, portraying officers as “terrorists,” and transitions to Dan Bongino's FBI resignation, exploring implications for internal leadership and Andrew Bailey's potential role. The hour blends politics, law enforcement, and media commentary with personal and local stories.

NEW THEME SONG VERSION - Thanks ClaytonThis episode has some of the best information I've put out there and how I compare stocks - but I used several AI agents to figure out strategies too! $TSLA vs. $RIVN and what do you think of $BAC? THESE SALES END SOON: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠TRENDSPIDER HOLIDAY SALE - Get 52 trainings for the next year at 68% off. Become a Trendspider master! ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠SEEKING ALPHA BUNDLE - ALPHA PICKS AND PREMIUM Save over $200⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Seeking Alpha Premium - FREE 7 day trial ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Alpha Picks - Save $100 ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Seeking Alpha Pro - for the Pros ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠EPISODE SUMMARY

Send us a textWe explore how foaming TPU, belt printers, and clinician-led automation can bring custom foot orthoses in-house, cut costs, and reduce lead times from weeks to hours. Fredrik from CADmed shares what actually works, what doesn't, and how remote design plus per-pair pricing changes the game.• why automation increases clinician control rather than removes it• carving versus printing tradeoffs on speed, finish, and cost• foaming TPU benefits including smoother surfaces and less waste• belt printers for continuous TPU production and better surface feel• FDM advantages over powder bed fusion for insoles• practical lattices and gradients without overengineering• decentralizing production to clinics for faster delivery• remote design, slicing, and printer control with per-pair pricing• scaling systems for reliability, repeatability, and low frictionSpecial thanks to Advanced 3D for sponsoring this episode.Support the show

Derniers modèles d’intelligence artificielle et outils comme Captain Safari et Anytalker qui transforment la production de contenu vidéo. Nous discutons également de l’état des travaux de George R. R. Martin sur « A Song of Ice and Fire » et des implications de l’IA dans la créativité littéraire. Défis liés à la fabrication de matériel informatique, notamment l’impact environnemental croissant des data centers.  Patreon YouTube Discord Modèles et idioties de la semaine Captain Safari, Any talker, VR Bench et Gen VIRE. L'IA se normalisera… doucement. Et en attendant, attention… Une moitié des nouveaux textes ne sont pas synthétiques. Est-ce encore trop ? C'est officiel, le DOGE a été une calamité et Elon en est toujours une. Georges RR Martin ne veut pas utiliser l'IA, donc c'est foutu. Explosion des datacenters : au moins les saumons seront contents. C'est Métal Dell sauvé par l'IA, pas par Windows 11. Partage de “talents” entre Intel et TSMC, Rapidus accélère. Partage de GPU avec la Chine. Et les TPU ? La DDR7 arrive ! Ça tombe bien, toutes les autres sont parties. Black Sky vous a à l'oeil et les magnétomètres quantiques. Bientôt un téléphone qui voit comme un dinosaure. C'est nickel : Google passe du Chrome à l'Aluminium. Y'a qu'une télé, c’est Telehack… Bonus : les beaux arts ASCII. Participants Une émission préparée par Guillaume Poggiaspalla Présenté par Guillaume Vendé

In this episode, we welcome Joel Elconin, co-host of PreMarket Prep and founder of the Stock Trader Network, to break down the market's steady rebound, shifting sector leadership, and what to watch as we approach year-end. Key Discussion Highlights Market Grind Higher The S&P and Nasdaq continue a slow climb toward all-time highs, with day-to-day rotation across tech, value, and retail. AI Trade Reset META cost cuts, Oracle's struggles, and Google's new TPU chips highlight a more competitive, and more selective, AI environment. Seasonality & Fed Expectations Santa Claus rally patterns, tax-loss selling, and shifting rate-cut odds are shaping year-end behavior. 2026 Themes Guests on Joel's shows continue to flag healthcare strength, selective value opportunities, and resilient retail trends. Stocks Mentioned: META, ORCL, GOOGL, CRM, BRK.A/B, AMZN   Click here to visit Joel's PreMarket Prep website Click here to visit the Stock Trader Network ----------------- For more market commentary & interview summaries, subscribe to our Substacks: The KE Report: https://kereport.substack.com/ Shad's resource market commentary: https://excelsiorprosperity.substack.com/ Investment disclaimer: This content is for informational and educational purposes only and does not constitute investment advice, an offer, or a solicitation to buy or sell any security. Investing in equities and commodities involves risk, including the possible loss of principal. Do your own research and consult a licensed financial advisor before making any investment decisions. Guests and hosts may own shares in companies mentioned.

寫程式不再是工程師的專利，而是人人都能掌握的新時代超能力！ 擁有 20 年程式開發經驗的 Mosky 老師，將為你精煉 1% 關鍵知識，搭配 AI 精準協作，你就能從零開始寫出可運行的程式，進而消滅瑣事、解放創意！ 一起用 AI 寫程式打造全新生活：https://pse.is/8c7jxr 輸入專屬折扣碼：MIULA，即可享有 $250 的折扣喔！ --- EP256. 再談 TPU 與 GPU、Cybercab 得到公道、學習與 AI 共同創作 | M觀點 --- (00:40) EP256 預告 (02:43) 業配時間：從零開始 AI 寫程式｜用 1% 關鍵知識消滅瑣事、解放創意 (09:09) 第一個話題：再談 TPU 與 GPU (33:53) 第二個話題：Cybercab 得到公道 (44:31) 第三個話題：學習與 AI 共同創作 --- M觀點資訊 --- 科技巨頭解碼: https://bit.ly/3koflbU M觀點 Telegram - https://t.me/miulaviewpoint M觀點 IG - https://www.instagram.com/miulaviewpoint/ M觀點Podcast - https://bit.ly/34fV7so M報: https://bit.ly/345gBbA M觀點YouTube頻道訂閱 https://bit.ly/2nxHnp9 M觀點粉絲團 https://www.facebook.com/miulaperspective/ 任何合作邀約請洽 miula@outlook.com -- Hosting provided by SoundOn

Welcome back to another episode of Upside at the EUVC Podcast, where ⁠Dan Bowyer⁠,⁠ Mads Jensen⁠ of ⁠SuperSeed⁠, ⁠Lomax Ward⁠ of ⁠Outsized Ventures⁠⁠⁠, and this week's special guest Robin Haak break down the real stories behind the headlines shaping European tech and venture.Robin joins us as the founder of Robin Capital, an early employee at SmartRecruiters, angel in 100+ companies, including eight unicorns, and one of the most active emerging GPs in Europe. He brings deep operator insight, especially into the German ecosystem, politics, and economy, which this episode leans heavily into.We cover everything from UK policy signals to German recession warnings, AI dominance to Europe's bureaucratic drag, the rise of solo GPs, and why the next decade of tech will be won or lost on energy availability more than anything else.What's covered:04:00 EU wants to restrict social media for minorsThe team debates the proposals to ban or limit social media for children under 16, the mental health case, and the tension between safety and overreach.06:00 Surveillance creep & messaging regulationRobin explains concerning drafts that would've allowed governments to read private messages. The group breaks down the slippery slope of “protect the children” legislation.10:00 UK Budget: surprisingly startup-friendlyDan and Lomax unpack EMI reforms, EIS/VCT clarity, and why the market reacted calmly. Signals of a more innovation-forward UK emerge.12:45 Lovable.ai's VAT scandal & Europe's compliance mazeA Swedish engineer's viral post on LinkedIn sparks a discussion on Europe's inconsistent VAT rules, compliance complexity, and whether hypergrowth and European regulation can co-exist.17:00 N26's long struggle with German regulatorsRobin, an early angel, offers an insider's view on the fintech's challenges—BaFin restrictions, governance issues, and the counterfactual: “Would N26 be worth €20B if it were French?”20:00 Germany's big macro problem: stagnation + overloadA brutally honest breakdown of the German economy: energy scarcity, migration overload, rising welfare costs, labor shortages, and political paralysis.28:00 Education, welfare, pensions & the cost structure crisisRobin explains why Germany's systems are buckling: the collapse of PISA scores, overloaded municipalities, and an economic model no longer supported by productivity.33:00 Nuclear shutdowns & Europe's AI energy deficitWhy Germany shut down its safest reactors, how it backfired, and why France and the Nordics will become the new AI infrastructure hubs.40:00 Startup ecosystem: the good, the bad, the bureaucraticFrom Munich's deep tech boom to notary nightmares, ESOP fixes, GmbH limitations, and how founders are learning to hack the system.55:00 The rise of Solo GPsThe team discusses the American roots, European trajectory, operator funds, fund-of-funds appetite, and why founders increasingly prefer solo GPs.01:00:00 AI CornerOpenAI's trillion-dollar capex future, Google's TPU resurgence, Anthropic momentum, Michael Burry shorting AI (and why it's misguided), and the geopolitics of compute.

Adam Kobeissi, Founder & Editor-in-Chief of The Kobeissi Letter, joins Zaid Admani to break down whether Google's TPU push is a real threat to Nvidia, which Big Tech giant is most likely to hit a $10 trillion valuation first, whether AI has effectively become too big to fail, and how labor-market weakness is pushing the Fed toward more rate cuts. Kobeissi explains why he thinks the S&P 500 will continue to go higher, and why investors will get left behind if they don't own assets.

ChatGPT fête ses 3 ans : retour sur une révolution technologique désormais ancrée dans nos usages. Google rebondit avec Gemini 3 et relance la bataille de l'IA face à OpenAI. Apple patauge. Les robots prolifèrent… mais sauront-ils prendre soin de leur hygiène ?Partenariat : Free Pro, le meilleur de Free pour les entreprisesAvec François Sorel (BFM Tech & Co) et Bruno Guglielminetti (moncarnet.com)ChatGPT a 3 ans : quel bilan pour l'IA générative ?Lancé en catimini le 30 novembre 2022, ChatGPT compte aujourd'hui plus de 800 millions d'utilisateurs hebdomadaires. Une progression fulgurante, marquée par l'intégration de l'interface vocale et la mémoire conversationnelle. François Sorel salue une nouvelle ergonomie technologique. Bruno Guglielminetti partage des usages concrets, bluffants, parfois presque… thérapeutiques. On s'accorde sur le risque de paresse cognitive qui nous guette. ChatGPT n'est pas un cerveau, c'est un outil — et il faut le manier avec recul.Google revient dans la course avec Gemini 3Après l'échec de Bard, Google a sorti Gemini 3, salué pour ses performances. Fait marquant : le modèle tourne en partie sur les puces maison TPU, marquant une rupture avec la dépendance aux GPU Nvidia. Selon certains analystes, Gemini pourrait concurrencer sérieusement OpenAI, au point d'inquiéter Sam Altman lui-même. L'IA, chez Google, devient un produit à part entière, maîtrisé de bout en bout.Apple : des chaussettes et des doutesApple semble en retrait. Malgré la promesse d'Apple Intelligence, peu de concrétisations sont visibles à ce stade. La firme fait davantage parler d'elle avec des produits… inattendus, comme la chaussette pour iPhone, objet statutaire vendu en édition limitée. Des rumeurs de licenciements et de départ de Tim Cook alimentent le sentiment d'une transition floue.Robots humanoïdes : promesses et absurditésTrois nouveaux robots humanoïdes ont été dévoilés ce mois-ci. La France, absente du hardware, pourrait tirer son épingle du jeu via le logiciel embarqué, à condition de légiférer en amont. Et puis, on se pose une question moins anecdotique qu'elle en a l'air : les robots se lavent-ils les mains ? -----------♥️ Soutien : https://mondenumerique.info/don

Ranjan Roy from Margins is back for our weekly discussion of the latest tech news. We cover: 1) Black Friday secrets 2) Google may sell its TPUs to Meta and financial institutions 3) Nvidia sends an antsy tweet 4) How does Google's TPU stack up next to NVIDIA's GPUs 5) Could Google package the TPU with cloud services? 6) NVIDIA responds to the criticism 7) HSBC on how much OpenAI needs to earn to cover its investments 8) Thinking about OpenAI's advertising business 9) ChatGPT users lose touch with reality 10) Ilya Sustkever's mysterious product and revenue plans 11) X reveals our locations --- Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. Want a discount for Big Technology on Substack + Discord? Here's 25% off for the first year: https://www.bigtechnology.com/subscribe?coupon=0843016b Questions? Feedback? Write to: bigtechnologypodcast@gmail.com Learn more about your ad choices. Visit megaphone.fm/adchoices

Pre-Thanksgiving chatter from the Lessins' Surf Shack: Jess, Brit, Dave, and Sam pinball from holiday-card automation to trillion-dollar AI geopolitics. Brit trades Minted for Canva+GPT, Jess admits to maintaining a 600-row address spreadsheet, Sam unveils Slow's Etiquette Book, and Dave still can't believe we can't pay in USDC. Don't worry this year's Thanksgiving edition will live up to its hype, the crew gets into the real stuff too: Google's TPU push vs. Nvidia's moat, Meta reportedly buying billions in TPUs, whether Google can shave 10% off Nvidia's revenue, and more.Chapters:06:45 The San Francisco consensus and Silicon Valley's real innovation marketing11:10 Elon Zuck and megaphone-powered distribution14:30 Why interface distribution will decide AI winner17:20 Memory isn't real lock-in switching between ChatGPT and Gemini21:40 OpenAI's identity crisis: Apple-style computer vs Meta-style attention26:30 Google complex vs OpenAI complex how the narrative flipped29:10 Google TPUs vs Nvidia Meta's rumored buying spree33:20 AI infrastructure economics depreciation CapEx margins36:00 Macro vs micro elections risk cycles 40:10 DOE's Genesis Mission and where the analogy breaks44:00 OpenAI's Jony Ive device timeline48:30 Why distribution still beats novelty53:15 Final takeaways: marketing distribution and business-model warsWe're also on ↓X: https://twitter.com/moreorlesspodInstagram: https://instagram.com/moreorlessYouTube: https://youtu.be/7BbWHm3KODwConnect with us here:1) Sam Lessin: https://x.com/lessin2) Dave Morin: https://x.com/davemorin3) Jessica Lessin: https://x.com/Jessicalessin4) Brit Morin: https://x.com/brit

Phil Rosen believes the recent tech sell-off has been overblown and creates new buying opportunities for investors. As for the TPU and GPU discussions surrounding Alphabet (GOOGL) and Nvidia (NVDA), he sees both companies continuing to prosper for the foreseeable future. Phil tells investors to brace for a Santa Claus rally as he sees many headwinds behind markets. He also sees a more dovish FOMC once Jerome Powell leaves his position as Fed chair. ======== Schwab Network ========Empowering every investor and trader, every market day.Subscribe to the Market Minute newsletter - https://schwabnetwork.com/subscribeDownload the iOS app - https://apps.apple.com/us/app/schwab-network/id1460719185Download the Amazon Fire Tv App - https://www.amazon.com/TD-Ameritrade-Network/dp/B07KRD76C7Watch on Sling - https://watch.sling.com/1/asset/191928615bd8d47686f94682aefaa007/watchWatch on Vizio - https://www.vizio.com/en/watchfreeplus-exploreWatch on DistroTV - https://www.distro.tv/live/schwab-network/Follow us on X – https://twitter.com/schwabnetworkFollow us on Facebook – https://www.facebook.com/schwabnetworkFollow us on LinkedIn - https://www.linkedin.com/company/schwab-network/About Schwab Network - https://schwabnetwork.com/about

一款AI（人工智慧）驅動的新聊天機器人異軍突起，牽動科技股的領頭羊之爭，也促使市場對於美股當紅炸子雞輝達（Nvidia）的需求產生新的疑慮。 文： 樂羽嘉 製作團隊：樂祈 ＊閱讀零時差，點這看全文

Jason Howell and Jeff Jarvis explore Google's Nano Banana Pro launch and Jason's unique sit-down with the leaders of Gemini 3, Nvidia's earnings amid AI bubble talks, Google's TPU deal with Meta, and Warner Music's Suno settlement. They also cover Trump's Genesis Mission, Gmail AI concerns, Anthropic Opus 4.5, and Character AI's age ban. Note: Time codes subject to change depending on dynamic ad insertion by the distributor. Chapters: 6:43 - Google launches Nano Banana Pro, an updated AI image generator powered by Gemini 3 8:21 - Nate B Jones example of how he used Nano Bana Pro 15:33 - Behind the scenes with Google's Gemini team - 3 insights that surprised me the most And these features come to NotebookLM 23:39 - Nvidia's Strong Results Show AI Fears Are Premature 24:30 - Google Further Encroaches on Nvidia's Turf With New AI Chip Push 26:03 - Nvidia's happy for Google 32:18 - What to know about Trump's order for the AI project ‘Genesis Mission' 38:07 - Google denies analyzing your emails for AI training - here's what happened 41:22 - Warner Music Group strikes ‘landmark' deal with Suno; settles copyright lawsuit against AI music generator 44:31 - Suno creators making 7m songs a day; trained on only $2k 51:54 - Anthropic introduces cheaper, more powerful, more efficient Opus 4.5 model 55:36 - Teens Are Saying Tearful Goodbyes to Their AI Companions 58:02 - Jony Ive and Sam Altman say they finally have an AI hardware prototype 58:53 - Sam Altman and Jony Ive have a 'lick' test for OpenAI's mysterious AI device, which they expect within the next 2 years 1:00:43 - OpenAI Partner Foxconn Plans Multibillion-Dollar US AI Push 1:01:34 - Meta chief AI scientist Yann LeCun is leaving to create his own startup 1:03:09 - Jony Ive and Sam Altman say they finally have an AI hardware prototype Learn more about your ad choices. Visit megaphone.fm/adchoices

The media narrative on Google (Alphabet) has flip-flopped again. Suddenly, Google TPUs are "killing" Nvidia, Gemini 3 is here, and the stock is soaring. But is AI dominance really the reason Google stock has doubled since April?In this episode, we dig past the headlines to uncover the real catalyst behind Google's recent stock performance—and it has less to do with the TPU vs. GPU debate and more to do with the clearing fog around major antitrust cases regarding Chrome and Android.We also break down Alphabet's massive $56B R&D spend, their aggressive AI data center CapEx, and why their impressive per-share profit growth makes them a potential "soft hedge" against Nvidia in your semiconductor portfolio. Plus, we touch on why Broadcom remains a key beneficiary of Google's custom silicon build-out.#GoogleStock #Alphabet #Nvidia #TPU #SemiConductors #ChipStockInvestor #AI #Antitrust #BroadcomJoin us on Discord with Semiconductor Insider, sign up on our website: www.chipstockinvestor.com/membershipSupercharge your analysis with AI! Get 15% of your membership with our special link here: https://fiscal.ai/csi/Sign Up For Our Newsletter: https://mailchi.mp/b1228c12f284/sign-up-landing-page-short-formChapters[00:00] The Media Flip-Flop on Google AI[01:00] Google's Profitability: EPS & Free Cash Flow Growth[02:22] The Real Catalyst: Antitrust Updates (Chrome & Android)[04:45] Analyzing the $56B R&D Budget: Money Well Spent?[06:20] Google as a "Soft Hedge" for Nvidia & Broadcom's Role[06:50] Conclusion & Upcoming Semis ReportsIf you found this video useful, please make sure to like and subscribe!*********************************************************Affiliate links that are sprinkled in throughout this video. If something catches your eye and you decide to buy it, we might earn a little coffee money. Thanks for helping us (Kasey) fuel our caffeine addiction!Content in this video is for general information or entertainment only and is not specific or individual investment advice. Forecasts and information presented may not develop as predicted and there is no guarantee any strategies presented will be successful. All investing involves risk, and you could lose some or all of your principal.Nick and Kasey own shares of Alphabet, Nvidia, Broadcom, Meta, Amazon

The battle for AI dominance just intensified. Google's latest press release unveiled major advancements in their TPU chips, raising a big question:

Kevin Green and Diane King Hall talk tech and technicals after Monday's market rally. Nvidia (NVDA) shares are under pressure as Meta Platforms (META) buys TPU chips made by Alphabet's Google (GOOGL). KG discusses the perceived battle between Nvidia and Google, as shares of Alphabet cross the $4T market cap threshold. KG is watching $170 to the downside on the NVDA chart as a level of support to watch and also underlines the potential breakout for another player in the space: Broadcom (AVGO). Later, KG provides his trading range to watch today with the S&P 500 (SPX) upside potential to 6750 and downside support near 6620.======== Schwab Network ========Empowering every investor and trader, every market day.Subscribe to the Market Minute newsletter - https://schwabnetwork.com/subscribeDownload the iOS app - https://apps.apple.com/us/app/schwab-network/id1460719185Download the Amazon Fire Tv App - https://www.amazon.com/TD-Ameritrade-Network/dp/B07KRD76C7Watch on Sling - https://watch.sling.com/1/asset/191928615bd8d47686f94682aefaa007/watchWatch on Vizio - https://www.vizio.com/en/watchfreeplus-exploreWatch on DistroTV - https://www.distro.tv/live/schwab-network/Follow us on X – / schwabnetwork Follow us on Facebook – / schwabnetwork Follow us on LinkedIn - / schwab-network About Schwab Network - https://schwabnetwork.com/about

Jessica Inskip turns to the tech trade for today's Big 3 with a focus on headlines between Mag 7 giants Alphabet (GOOGL) and Nvidia (NVDA). She sees significant upside for the Google parent company after acquiring a TPU deal with Meta Platforms (META), though she urges investors not to discount Nvidia and its wide A.I. moat. As for Oracle (ORCL), Jessica expects high reward to balance out high risk in its A.I. spending. Rick Ducat highlights the bearish and bullish trends he sees in the stock charts. ======== Schwab Network ========Empowering every investor and trader, every market day.Subscribe to the Market Minute newsletter - https://schwabnetwork.com/subscribeDownload the iOS app - https://apps.apple.com/us/app/schwab-network/id1460719185Download the Amazon Fire Tv App - https://www.amazon.com/TD-Ameritrade-Network/dp/B07KRD76C7Watch on Sling - https://watch.sling.com/1/asset/191928615bd8d47686f94682aefaa007/watchWatch on Vizio - https://www.vizio.com/en/watchfreeplus-exploreWatch on DistroTV - https://www.distro.tv/live/schwab-network/Follow us on X – / schwabnetwork Follow us on Facebook – / schwabnetwork Follow us on LinkedIn - / schwab-network About Schwab Network - https://schwabnetwork.com/about

The Information's Anita Ramaswamy talks with Erin Woo about Google's new strategy to compete with Nvidia by pitching its TPU chips to major companies like Meta and financial institutions for use in their own data centers. We also talk with AI reporter Stephanie Palazzolo about Sam Altman's concern over Google's pretraining advantage with Gemini 3. Next, we get into rising debt funding for Oracle's data center build outs with Finance reporter Miles Kruppa, and Ran Ben-Tzur discusses the IPO market heading into 2026. Finally, we speak with ModelML CEO Chaz Englander about their $75 million Series A round and the use of AI to automate tasks in financial services.Articles discussed on this episode:https://www.theinformation.com/articles/google-encroaches-nvidias-turf-new-ai-chip-pushhttps://www.theinformation.com/articles/oracle-linked-borrowing-binge-worries-lendersTITV airs on YouTube, X and LinkedIn at 10AM PT / 1PM ET. Or check us out wherever you get your podcasts.Subscribe to: - The Information on YouTube: https://www.youtube.com/@theinformation- The Information: https://www.theinformation.com/subscribe_hSign up for the AI Agenda newsletter: https://www.theinformation.com/features/ai-agenda

Welcome back to another episode of Upside at the EUVC Podcast, where ⁠Dan Bowyer⁠,⁠ Mads Jensen⁠ of ⁠SuperSeed⁠, ⁠Lomax Ward⁠ of ⁠Outsized Ventures⁠⁠⁠ dissect the stories reshaping European venture, from Helsinki's Slush takeover to China's rising leverage, TPU vs GPU battles, the UK's AI money wave, and why immigrants found half the unicorns in the Western world.This week's episode ranges from Germany's €35B space ambitions to Meta's TPU dealmaking, from cookie law rollbacks to Lithuania's secondhand unicorn, all culminating in one conclusion: Europe's window for action is open, but narrowing.

(0:00) Bestie intros LIVE from The Venetian Las Vegas (1:13) Epstein Files breakdown (10:06) Biggest Epstein questions: where did his money come from? (14:44) Tether's booming business (23:50) Michael Burry vs. Friedberg, Nvidia's blowout quarter and risks for 2026 (35:25) Google's Gemini 3 and TPU breakthrough (42:51) Investing your own money vs. LP capital, why Friedberg returned as a CEO (48:57) Alan Keating joins the show to talk poker strategy, thriving in chaos, risk psychology Special thanks to The Venetian Las Vegas for hosting us!: https://x.com/VenetianVegas Join us at the All-In Holiday Spectacular!: https://allin.com/events Follow the besties: https://x.com/chamath https://x.com/Jason https://x.com/DavidSacks https://x.com/friedberg Follow on X: https://x.com/theallinpod Follow on Instagram: https://www.instagram.com/theallinpod Follow on TikTok: https://www.tiktok.com/@theallinpod Follow on LinkedIn: https://www.linkedin.com/company/allinpod Intro Music Credit: https://rb.gy/tppkzl https://x.com/yung_spielburg Intro Video Credit: https://x.com/TheZachEffect Referenced in the show: https://x.com/RepClayHiggins/status/1990868089056219267 https://x.com/michaeljburry/status/1991289193037746579 https://polymarket.com/event/which-company-has-best-ai-model-end-of-2025 https://x.com/Similarweb/status/1988879389992386897 https://x.com/PokerGO/status/1987406318832132256

This week we dive into everything from viral content and controversial sports calls to tech upgrades, shop hacks, and the frustrations of Amazon!Tech and Tool Talk:The iPhone Camera Bump Solution- We solve the biggest problem with modern phones: the stupid camera bump! We pitch solutions to make your phone flat—and give you a 25% larger batteryMac Modernization- Zac reveals his latest crazy electronics project: building a fully modern Mac Mini computer inside an old, iconic iMac G3 shell, complete with an OLED screen and sound system upgradeSharpening Success- Deric details his fantastic experience sharpening his chisels with the Zen-Wu sharpening system and getting a mirror finish that could shave hair off your armsThe TPU Fail- Eric shares the full story of his failed attempt to create a grippy, 3D-printed TPU clamping block to stop miter joints from slippingViral Strategies- We analyze viral content strategies, from a guy whose social media is dedicated to buying and returning the same anvil, to an idea that could give us the next woodworking hitSports Controversy: The guys debate AI umpires in baseball and the arbitrary nature of "traveling" calls in the NBATopics: Woodworking, clamping blocks, 3D printing, flexible filament, tech projects, phone design, hand tool sharpening, business practicesGot a question that you want us to answer? Send us an email at ⁠⁠offthecutpodcast@gmail.com⁠ -------------------------AftershowGet access to the aftershow and unlock tons of cool perks over on Patreon-⁠⁠⁠⁠⁠https://www.patreon.com/offthecutpodcast⁠  ⁠⁠⁠-------------------------Hang Out with UsWatch the live stream of the podcast on YouTube!⁠⁠⁠⁠⁠https://www.youtube.com/channel/UCcRJPIp6OaffQtvCZ2AtWWQ⁠⁠⁠⁠⁠-------------------------Pick Up Some Merch!Off The Cut Podcast- ⁠⁠⁠⁠⁠https://www.spencleydesignco.com⁠⁠⁠⁠⁠   -------------------------Follow ZacInstagram - ⁠⁠⁠⁠⁠https://www.instagram.com/zacbuilds⁠ YouTube - ⁠⁠⁠⁠⁠https://www.youtube.com/c/@ZacBuilds⁠⁠⁠⁠⁠ TikTok - ⁠⁠⁠⁠⁠https://www.tiktok.com/@zacbuilds⁠⁠⁠⁠  -------------------------Follow EricInstagram - ⁠⁠⁠⁠⁠https://www.instagram.com/spencleydesignco⁠⁠⁠⁠⁠ YouTube - ⁠⁠⁠⁠⁠https://youtube.com/@spencleydesignco⁠⁠⁠⁠ TikTok - ⁠⁠⁠⁠⁠https://www.tiktok.com/@spencleydesignco⁠⁠  ⁠-------------------------Follow DericInstagram/YouTube/TikTok @PecanTreeDesign ⁠⁠⁠⁠⁠https://linktr.ee/pecantreedesign⁠⁠⁠⁠⁠ ---------------------------This episode is proudly sponsored by:KM Tools - Check out everything they have to offer at ⁠⁠⁠kmtools.com/SPENCLEYDESIGNCO⁠ ⁠WTB Woodworking⁠ - Check out the giveaway over at:⁠https://www.wtbwoodworking.com/giveaway⁠ Gorilla Glue - Built By You; Backed By Gorilla www.gorillatough.com Interested in starting your own podcast? Check out Streamyard: ⁠⁠⁠⁠⁠https://streamyard.com/pal/c/5926541443858432⁠  ⁠⁠ ⁠#Woodworking #DIY #3DPrinting #Maker #ContentCreation #YouTuber #OffTheCutPodcast #Sponsored #KMTools #WTBWoodworking #GorillaGlue 

Josh tackles why we're still pinning race numbers instead of printing them on jerseys, then dives into the tech story behind Silca's translucent TPU tubes. We talk about real-world marginal gains for everyday riders, and Fatty's recommendations for indoor training setups. Last but not least, you do NOT want to miss Josh's unfiltered take on the UCI–SRAM 10-tooth cog controversy.

Microsoft Reporter Aaron Holmes talks with TITV Host Akash Pasricha about the $38 billion AWS-OpenAI compute deal and the practical challenges of deploying AI agents in the enterprise. We also talk with StarCloud Co-Founder & CEO Philip Johnston about launching an NVIDIA H100 GPU into space, its core cooling technology, and their business model. Then, Google Cloud VP Oliver Parker discusses their AI go-to-market strategy, the TPU platform, and their strong earnings results, and we get into the enterprise AI adoption hurdles and ServiceNow's heterogeneous chip stack with Vice Chairman Nick Tzitzon. Finally, Tubi CEO Anjali Sud explains how the free, ad-supported streaming service reached profitability and their 'Free Forever' philosophy.Articles discussed on this episode:https://www.theinformation.com/briefings/openai-aws-sign-38-billion-cloud-dealhttps://www.theinformation.com/articles/anthropic-aws-give-customers-ai-agents-helping-handTITV airs on YouTube, X and LinkedIn at 10AM PT / 1PM ET. Or check us out wherever you get your podcasts.Subscribe to: - The Information on YouTube: https://www.youtube.com/@theinformation4080/?sub_confirmation=1- The Information: https://www.theinformation.com/subscribe_hSign up for the AI Agenda newsletter: https://www.theinformation.com/features/ai-agenda

出國上網，不用再那麼麻煩了 由 NordVPN 所推出的 Saily eSIM 服務，真的方便好用 下載 Saily eSIM 的 APP，在裡面購買並且啟用，eSIM 就可以開始作用 超過 200 個地區可使用，還可以幫你追蹤網路使用量 立刻點及專屬連結下載 Saily APP，並在結帳時使用優惠代碼 [miula]，立即享有專屬 eSIM 方案 85 折優惠！ M觀點 X Saily eSIM - https://saily.com/miula --- EP246. 谷歌會外賣 TPU 嗎、川習即將要見面、稀土股的投資思考 | M觀點 --- (00:40) EP246 預告 (02:58) 業配時間： Saily eSIM (05:34) 第一個話題：谷歌會外賣 TPU 嗎 (30:03) 第二個話題：川習即將要見面 (43:48) 第三個話題：稀土股的投資思考 --- M觀點資訊 --- 科技巨頭解碼: https://bit.ly/3koflbU M觀點 Telegram - https://t.me/miulaviewpoint M觀點 IG - https://www.instagram.com/miulaviewpoint/ M觀點Podcast - https://bit.ly/34fV7so M報: https://bit.ly/345gBbA M觀點YouTube頻道訂閱 https://bit.ly/2nxHnp9 M觀點粉絲團 https://www.facebook.com/miulaperspective/ 任何合作邀約請洽 miula@outlook.com -- Hosting provided by SoundOn

AI systems just got a huge context boost. Anthropic adds memory to Claude, OpenAI launches “Company Knowledge” to connect ChatGPT directly to enterprise data, and Microsoft debuts long-term memory and shared context in Copilot. Plus, Oracle's record $38B debt deal to fund AI infrastructure, Google's massive TPU expansion with Anthropic, and a real-world success story showing what vibe coding can do.Brought to you by:KPMG – Discover how AI is transforming possibility into reality. Tune into the new KPMG 'You Can with AI' podcast and unlock insights that will inform smarter decisions inside your enterprise. Listen now and start shaping your future with every episode. ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.kpmg.us/AIpodcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠AssemblyAI - The best way to build Voice AI apps - ⁠https://www.assemblyai.com/brief⁠Blitzy.com - Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://blitzy.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ to build enterprise software in days, not months Robots & Pencils - Cloud-native AI solutions that power results ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://robotsandpencils.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠The Agent Readiness Audit from Superintelligent - Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://besuper.ai/ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠to request your company's agent readiness score.The AI Daily Brief helps you understand the most important news and discussions in AI. Subscribe to the podcast version of The AI Daily Brief wherever you listen: https://pod.link/1680633614Interested in sponsoring the show? sponsors@aidailybrief.ai

The gang is back with a spicy opener: Brit calls out Dave for dismissing her “AI-assisted” doc as work slop, while Sam argues lowbrow culture now imitates machines and highbrow culture resists them. We break down OpenAI's early, hallucination-prone Atlas browser and debate whether it's a real agent platform, just memory capture, or total $$$ burn. Sora hype is cooling as TPUs rise—with Jessica noting how The Information's TPU scoop even nudged NVIDIA to move faster. OpenAI's compute could approach ~$45B/year at ~3 GW, forcing new financing models and tough research vs GTM trade-offs—Sam calls it Pascal's Wager for Big Tech. We close on AWS outages, crypto leverage swings, and why “swag is dead,” replaced by bespoke gifting as the new status move. Just another week of your favorite Technology podcast with the Morins and Lessins.Chapters:00:38 — Gang is back: notes on last episode and Yasso bars01:20 — Revisiting last week's “AI Work Slop”: Brit vs. Dave 03:49 — Culture vs. LLMs: highbrow resists, lowbrow imitates07:16 — Bot etiquette: Bot joining Zoom calls and writing thank you letters12:16 — OpenAI Atlas: early tests, hallucinations & agency potential17:29 — Sora hype cools: the consumer AI novelty cycle19:47 — The TPU plot twist: NVIDIA gets pushed to move faster22:40 — $45B compute future: Pascal's Wager for Big Tech29:05 — AWS outages: one-offs vs systemic risk33:29 — The X Game: Mr.Beast, “manual slop,” and the screenshot era43:38 — Swag is dead: bespoke gifting = new status flex53:43 — Sports, socials & 2026 predictionsWe're also on ↓X: https://twitter.com/moreorlesspodInstagram: https://instagram.com/moreorlessYouTube: https://youtu.be/VDYdIoHe4XAConnect with us here:1) Sam Lessin: https://x.com/lessin2) Dave Morin: https://x.com/davemorin3) Jessica Lessin: https://x.com/Jessicalessin4) Brit Morin: https://x.com/brit

In this week's roundup, we discuss the pardon of convicted Binance founder, Changpeng Zhao, the surprising return of a Clippy-inspired "Mico" as Microsoft leans into more personable AI assistants, and OpenAI's acquisition of Sky as the agentic desktop race accelerates. We also cover Anthropic's planned TPU expansion with Google to support enterprise demand and frontier-model training, and we explore reports of the Trump administration eyeing equity stakes in U.S. quantum computing firms. The Chart of the Week highlights current crypto adoption trends. Remember to Stay Current! To learn more, visit us on the web at https://www.morgancreekcap.com/morgan-creek-digital/. To speak to a team member or sign up for additional content, please email mcdigital@morgancreekcap.com Legal Disclaimer This podcast is for informational purposes only and should not be construed as investment advice or a solicitation for the sale of any security, advisory, or other service. Investments related to the themes and ideas discussed may be owned by funds managed by the host and podcast guests. Any conflicts mentioned by the host are subject to change. Listeners should consult their personal financial advisors before making any investment decisions.