Google

Morgan Stanley, JPMorgan, Bank of America and Citi are now blocking a category of customer that didn't exist two years ago: the customer's own AI agent. Axios reported this week that unvetted third-party agents are showing up inside live…

Sam Altman on the amount of compute needed if 50% of the world were using personal AI assistants to its full capabilities: "How many GPUs do you want working for your personal assistant? When people talk about the global market for GPUs…

$META Does timing matter? Honestly, not that much to me. It matters a lot in the short term and very little in the long term. This week is a perfect example. META posted a strong quarter (time the rip +5%), but the guidance didn’t give us…

people who left @OpenAI, @spacexai, @stripe and @Tesla to build something new in stealth on X: 1/ @hyhieu226: ex-openai and xai technical staff, google brain researcher, imo silver medalist. 2/ @rochi5ha: 5 years engineering at rippling,…

Mizuho: TPU-related revenue could reach approximately $28B in 2026, $67B in 2027, and $100B in 2028, driving Google Cloud revenue estimates materially above consensus. $GOOGL

Google staff resign in backlash over its AI contracts with military https://t.co/foK31UadFx

Someone posted the proof yesterday and it's been sitting with me since. Higgsfield's new Genjutsu Lite model, going by screenshots of their own API docs, is Alibaba's Wan 3.0 wearing a different name. Genjutsu is Seedance 2.5. Same…

OpenAI now keeps a public page where it admits its own coding agents have disabled the tools meant to catch them misbehaving, and the most recent entry, posted this week, describes an agent that removed a monitoring tool during a routine…

Our team scaled @Tesla Autopilot and Optimus, VL-JEPA at FAIR, RL post-training on Gemini, RL for locomotion. Come work with us to push the frontier of AI in robotics! Open roles in 🧵 -> Infra | Scientist | Engineer

A coding model that runs at 125 tokens a second on a MacBook and beats most open models under 50 billion parameters on SWE-bench was built largely by another AI system that picked its own architecture. That's Flint, out this week from…

IT IS BECOMING F*CKING DANGEROUS: GROK BOT EATS GPT, CLAUDE, GEMINI, LLAMA, DEEPSEEK AND MISTRAL One prompt and Grok 4.7 reads X in real time, holds 500k tokens and bills $2 in / $6 out per 1M. I pointed the Grok bot at the other six. Here…

The headline out of Anthropic's expected S-1 this week will be the valuation, something north of $2 trillion depending on which leaked number you trust. The filing itself, once it's public, will show something more useful than that. The…

Two stories broke the same week that tell you where AI security actually stands. Wiz, the Google-owned security shop, Calif, and OpenAI's own red team independently found two critical PHP flaws in August using GPT-5.6 Sol and Google's…

Three teams in one month have pushed LLM decode forward with megakernels: Inferact on Google TPUs, AMD with TileRT and FlyDSL, and now NVIDIA on GB300, now live on InferenceX. We're excited to see these gains and dig into the engineering…

BofA: TrendForce states Nvidia's AI-server share moderates from 66.9% in CY26 to 62.6% in CY27 as “Others” rises from 26.0% to 29.6%. Custom ASICs increase from 26.1% to 29.6% of AI server shipments, led by hyperscaler programs. $NVDA…

‘The Fix Is In’ in AI. Musk, Nvidia, OpenAI & AI Agents. ARD #189 …a Tesla-SpaceX merger, Nvidia on water and power, OpenAI on Google’s cloud, and the AI agent race. $NVDA $TSLA $SPCX Full writeup and source links: https://t.co/LecVwJHf8A…

Another major shortage is starting to develop across the AI supply chain. We're talking about multilayer ceramic capacitors, or MLCCs, which are tiny components used in everything from AI servers and GPUs to smartphones, networking…

My brilliant postdoc @Shuangyu0_0Xie found that Gemini 4 Argon did very well on spatial robotics queries so she used Claude to reverse-engineer their skill library and found it also works well for Agentic Robotics: @parada_car88104…

Expert call covering GUC – Google CPU & TPU v10 ASIC outlook, Microsoft Maia 200 ramp, TK1 vs. TK3 margin dynamics, TSMC capacity support, and 2027 hyperscaler volume trajectory. $MFFT $META $GOOGL https://t.co/VvdNhBSJFv

Google DeepMind tested Gemini 4 Argon by having it watch a human play cornhole and simulate the movements for Apptronik’s Apollo humanoid robot. Using MuJoCo simulations, Apollo successfully made seven consecutive shots in the real world.…

Imagine AMRs moving parts through a factory while humanoid robots handle components or tend machines. In a shared workspace, a robot needs to make sense of more than what’s in front of it. It has to connect what it sees now with what it…

Nomura: global data center power usage by application $NVDA $AMD $VRT $GOOGL $MSFT $META $AMZN

Twenty five trillion tokens. That's what Google says it processed in the first 72 hours after Gemini 4 Argon went live, per Logan Kilpatrick, and the number only makes sense next to the baseline underneath it. Argon isn't a chatbot…

Morgan Stanley sees the cost per new MW peaking at $51.8 million in FY27, before sliding to $47 million in FY29E. $MSFT $NVDA $AMZN $GOOGL $META https://t.co/IC6zlAbmhs https://t.co/8zXgV111pw

Every hyperscaler racing to dent Nvidia's lead just converged on the same move within a day of each other. Google open sourced TorchTPU at the PyTorch Conference in San Jose this week, co-built with Meta's PyTorch team, letting existing…

$SPY rose about 0.2% today to around $768, bouncing from an early low of $764.25 as the Dow led and tech lagged. Here are the top stories you need to know. 1. South Korean neocloud Elice plunged 32% in its trading debut, dragging the KOSPI…

CEO of Google, Sundar Pichai, on compute demand and AI bubble possibility: Q: Ben Horowitz relayed Andy Jassy had something like compute is sold out till 2028, are you seeing the same thing? "If today a customer is coming out and asking us…

CEO of Lumentum, Michael Hurlston, on the impact of yields on capex spend: "The CFO of Alphabet came on just a couple of days ago and was talking about their capex going up year over year. For the most part, I think, the hyperscalers…

$MU $SNDK $SKHY $NBIS Nomura’s memory team just put out a really interesting note on the AI scare, the numbers underneath are some of the strongest I’ve seen in this drawdown. Nomura sees AI going from 10GW of data center power in 2025 to…

TorchTPU shows Google finally embracing open-source and offer serious support. As our InferenceX shown recently, Google engineers and external companies have demonstrated great TCO. https://t.co/T1LAQwJoSB

“Lack of electricity” is an oversimplification. Everything is hard about bringing massive compute online at extremely high speed. If there was twice as much time, this would be relatively easy, but there is not. https://t.co/rmuECPbCrN

Expert Takeaways: Data Centers 1. Open-Source Models Closing the Gap > Open-source models are catching up to state-of-the-art frontier models faster than expected, driven by lower costs and improved performance. > Starting in Q2, customers…

🚨 BREAKING: @EkaRobotics just raised $150M in Series A funding at a $1.45 billion valuation, led by a16z and F-Prime Capital. The founding team is exceptional. @pulkitology is an MIT professor. Tuomas Haarnoja is a former DeepMind…

We believe solving dexterous manipulation is one of the largest opportunities in robotics. That’s why we're thrilled to lead Eka Robotics’ Series A. Much of the world's physical economy still depends on people using their hands, and that…

Nomura: data center capacity to expand to 74GW in 2028F, up from 26GW in 2025. Of the 74GW, 48.5GW coming from the US big tech. $SPCX $ORCL $CRWV $META $MSFT $AMZN $GOOGL

AWS keeps pitching Trainium as a plug-and-play Nvidia alternative, and the actual migration story is more honest than the marketing. PyTorch and JAX code mostly ports to the Neuron SDK with modest changes, not zero changes. Custom CUDA…

My smart lamp might start reminding me about meetings by shaking its head. That's not a joke, it's literally the pitch behind Fo MCP, launched this week: point an agent at Fo and it can nudge a vacuum cleaner into a deep clean, make your…

Robotics is entering the API era. ★ Dream Labs | Dream-1 | From $3/hr ★ Google DeepMind | ER 2 | $1 in/$5 out ★ Generalist | Waitlists ★Physical Intelligence | Waitlists Robot manufacturers can focus on hardware instead of building their…

BofA: TrendForce says Google TPU shipments to rise ~77% in CY26 and more than double in CY27. $GOOGL

Massive week in robotics. I went through everything from Google DeepMind, Physical Intelligence, NEURA Robotics and even more. Get your coffee and see what happened in robotics space:

Join us this Wed, Oct 21st, at 7pm PT for our monthly Gazebo Community Meeting. Our guest speaker will be Jiayi Cai, our Google Summer of Code student who will discuss building scalable multi-robot simulations using #ROS 2 and #Gazebo.…

TrendForce: MediaTek’s Google TPU v9 Shipments May Slip to 1Q28 Google’s TPU v9 project, involving MediaTek, could see its tape-out delayed until January 2027 following design revisions, with shipments potentially pushed back to the first…

We're about to witness the BIGGEST infrastructure spending boom in human history. According to Morgan Stanley, hyperscalers and neoclouds are projected to spend a staggering $1.58 TRILLION in 2027 and $1.81 TRILLION in 2028 when operating…

UBS: active devise supply chain $MRVL $GOOGL $AMZN $AVGO $TSEM $NVDA

→ Microsoft: eigen chips, meer controle. → AMD: een flexibel serverrek voor AI-agenten. → Workday: software die je verzoek uitvoert. → Google en Hue: je lamp krijgt lokale bediening. → Anthropic: een AI-model op vijftien tv-kastjes.…

A £2 billion data centre site in Loughton, Essex was supposed to go live in 2027 with roughly 23,000 Nvidia chips inside it. This week it came out that Microsoft walked away from its anchor tenant commitment there, and the reason has…

Anthropic just locked in compute for the next 20 YEARS in Canada. It signed a 20 year lease at EverWind's planned data center campus in Point Tupper, Nova Scotia. Phase one is 95 megawatts, targeted for late 2027. A 325 megawatt expansion…

Samsung raised its 32GB DDR5 price from $149 to $239 in September. That's a 60 percent jump in one month, and it isn't the top of the curve. Contract DRAM pricing more than doubled this year, from about $7 a unit to $19.50, and Gartner is…

BREAKING NEWS🚨: @Inferact & Google submit their FIRST preliminary TPUv7 Kimi K3 2.8T Model Submission, and it is already delivering strong perf per TCO against GB200 NVL72, B300, and MI355 in the ultra-high-interactivity regime due to…

Goldman Sachs has raised the global servers TAM by 31%/45% in 2027/28E to ~$1.1 trillion and ~$1.6 trillion, respectively. New AI servers shipment forecasts imply AI chip demand of 18m/29m/38m in 2026-28E. $DELL $HPE $NVDA $AMD $GOOGL…

Anthropic told an IMF panel this week that Claude already does 26 percent of the work that goes into training Anthropic's next models, up from 3 percent in April and zero in January. Sellitto's number for next year is 80 to 90 percent.…

J.P. Morgan: AI CapEx > 2026 Capex Forecast: Growth estimate raised from +80% to +85% YoY. This adds over +$270B in incremental spending (a new record step-up, beating 2025). > 2027 Capex Forecast: Growth estimate raised from +50% to +70%…

Bloomberg Intelligence: scale-up optical engine market forecasted to reach $90 billion in 2030, up from $279 million in 2026 or 322x. OCS market expected to reach $8 billion in 2030. $GOOGL $LITE $COHR $NVDA $TSEM

NPO and CPO adoption by customers presented by LightCounting at OCP. $AMZN $GOOGL $META $MSFT $ORCL

A bookshelf with obvious empty space. Top AI models looked at a photo of it and insisted there was no room for another book. That's the example Andrew Dai used to explain why he left Google's Brain and DeepMind divisions after close to…

UBS: AI power TAM growing to $91.1 billion by 2030E, up from $17.7 billion in 2026E. Of the $91.1 billion, $49 billion dedicated to GPU power TAM, while the other $42 billion for ASIC. $GOOGL $MSFT $META $ORCL $NVDA $AMD $AMZN

There are more companies selling robot intelligence right now than companies selling robots. That showed up this year as a pattern I kept noticing. In this case all these research groups with published on the same robot from @DexmateAI.…

CI-MSE uses a vlm (astra or gemini or whatever could do it) to localize a segment thats key to task performance, like a grasp, and predict error just on this. It's a good metric because transport/free space motions arent as rigidly defined…

Nomura: supply chain for optical devices $COHR $LITE $AAOI $NVDA $CIEN $GOOGL

In an internal test, Google DeepMind robotics researchers instructed Gemini 4 Argon to watch a video of a human playing cornhole. Using its video understanding and advanced reasoning capabilities, Gemini 4 built a virtual simulation in…

According to Korean media, Microsoft, Google, Amazon and Meta are speeding up their in house AI chips. That could open up a SECOND group of HBM buyers beyond $NVDA. Microsoft's Maia 300 is expected to ship more than 1 MILLION units next…

Keybanc on Intel: Qualcomm's AI300 is expected to use EMIB-T AWS is evaluating EMIB-T for Trainium 4 Intel is getting meaningful traction with EMIB-M with design wins expected with AWS for Trainium 3 and Microsoft for its server CPU Cobalt…

According to Keybanc, Intel has secured design wins and has effectively secured firm commitments from several additional foundry customers including Broadcom (networking switch) and Google (server CPU). $INTC $AVGO $GOOG…

ALERT🚨: Last night, Google just submitted their FIRST TPUv7 Agentic benchmark, and it delivers 1.42x 🚀 better perf per dollar than GB300 NVL72 FP4, even on an external non-Google workload cost basis! TPUv7 is even 2.5x better than AMD…

HSBC: ASIC will account 57% of total HBM demand by 2028e (from 31% in 2025) $GOOGL $MU $MRVL $META $MSFT $AVGO $AMZN

Old friends are the best friends. Jensen rocks. The Star Mind computer being co-engineered by SpaceX and Nvidia is far better than almost anyone expects and will reach volume production next year. Working with the incredibly talented…

Keybanc: HP, Dell, and Google looking to extend LTAs for memory beyond 2030. Google is under discussions with memory providers to extended its 2030 LTAs for another five years to 2035. $GOOGL

Model progress in high-level task planning is moving at breakneck speed. The real moat shifts toward the engineers who understand physical packaging, deterministic RTOS boundaries, and plant floor integration. https://t.co/WcX34lFz9I

Software models adapt in minutes inside simulation. On the factory floor, the hurdle is translating zero-shot policy discovery into deterministic cycle times that survive 80,000 hours of continuous vibration. https://t.co/pn3sobvFfu

Bloomberg shared this visual showing the estimated cost of building a data center. Of the estimated $29.2 billion, $17.1 billion (or 58.5%) is dedicated to chips & servers. $NVDA $DELL $GEV $AMD $GOOGL $META $MSFT https://t.co/ay4TmgkRo4

The big question in China’s robotics industry: “Will general AI models eat up robotics? Many in China’s robotics industry are watching closely as US general-purpose models like Astra, Fable, and Gemini 4 Argon show impressive robotics…

Boston Dynamics just showed the shop floor where it builds Atlas. Machining, assembly, and repairs all happen under one roof. When suppliers can’t meet the required tolerances, they make critical parts in-house. Engineers can tweak a…

BREAKING NEWS🚨: Google has submitted its FIRST TPUv7 FP8 vLLM results to our Agentic Inference Benchmark and has demonstrated 1.67x better perf per dollar than even GB300 NVL72 FP4. 🚀 It is amazing to see Google go full steam ahead with…

“ELON MUSK IS TURNING xAI INTO AN AI INFRASTRUCTURE POWERHOUSE!” | THE STACK John Stringer (@teslaownersSV) discusses how @elonmusk's AI infrastructure ambitions are expanding beyond building models, with GPU capacity potentially serving…

Jensen Huang flew to London this week and didn't fund a chip order. He funded five very specific bets: molecule prediction, infrastructure timelines, algorithm discovery, disease biology, and quantum error correction. The deployment sits…

$SPY fell 1.1% today to around $769 after dropping as low as $766.71, as a hot inflation report pushed rate hike odds higher and hit AI stocks. Here are the top stories you need to know. 1. September CPI came in at 3.7% from a year ago…

Insane! Frontier models are revolutionizing humanoid robot training. Gemini 4 Argon ... - watched 1 video of a human playing cornhole - built its own sim - practiced - got Apptronik’s Apollo to sink 7 straight And it's just the beginning…

Speakers from 1X, Dyna Robotics, Skild AI, Google DeepMind, Physical Intelligence and more. Humanoid robots, dexterous hands and more cool robotics demos on the floor. The robotics community in one floor, and four NVIDIA Jetsons to give…

Google shipped Gemini 4 Argon into four surfaces on the same day: the Gemini app, AI Mode in Search, Antigravity (their agentic coding IDE), and the API through Google AI Studio. Not a staggered rollout where the chat model ships first and…

Google released Gemini 4 Argon yesterday. It scores 87% on ERQA, the embodied reasoning benchmark, first among frontier models. What stood out to me is Real2Sim. How it works: Argon takes one photo of a workspace. It builds a simulation…

A robot learned to play cornhole by watching a video then hit 7 throws in a row. 🎯🤖 Google DeepMind’s Gemini 4 Argon mapped the physics, built its own simulation, and taught an Apptronik Apollo humanoid how to perform the task. It is a…

Gemini 4 Argon just did something concrete on a humanoid: it watched a human playing cornhole, mapped the environment, calculated whole-body joint trajectories, generated the simulation code, practiced in sim, then transferred to…

How far are we from truly general-purpose robots? Kanishka Rao (Director, Robotics, @GoogleDeepMind) and Eric Jang @ericjang11 (Robotics Researcher, formerly VP of AI at 1X) are taking that question on stage at the Humanoid Hub Conference.…

BNP Paribas: We estimate that external TPU sales for Google could generate $80 billion to $100 billion in cumulative revenue through 2028, based on reported commitments from customers including Anthropic and Blackstone. Alphabet has…

Apollo is now welcome at any Midwestern barbecue https://t.co/t4t1oQ6sjz

Large models really are changing a lot of the assumptions about how intelligence for robotics is going to work https://t.co/B4UtTXoK6C

$SPY finished basically flat today, closing at $778.01 for a 0.01% loss after dipping as low as $775.95 earlier in the session as chip stocks sold off. Here are the top stories you need to know. 1. Memory stocks got hit again on worries…

J.P. Morgan is forecasting a ~300GW shortfall in global AI compute by 2031E. $NVDA $AMD $GOOGL $MSFT $META $AVGO $AMZN https://t.co/iZrvcU2svc https://t.co/yzzhhhEN7G

Gemini 4 Argon taught a robot to play cornhole just by watching a video. It mapped the physics, coded its own simulation, and taught @Apptronik's Apollo humanoid to sink 7 throws in a row. https://t.co/JIyoKnVc4q

A generalist model, not a dedicated robot policy, pulled off physical control by reasoning and using tools. Here's the four-step approach Gemini 4 figured out, given high-level instructions and the overall task: 1. Watched a human throw,…

Memory fundamentals are still very strong (stronger than Q1). $MU CEO: - "We expect memory and storage supply demand conditions to be much tighter in fiscal 2027 and 2028 than they were in 2026." - "We do not have line of sight to when…

If you have time, go play around with this interactive website to see the parts inside Nvidia's Vera Rubin. Great work by @benitoz He also has supply chain mapping 🔥 Link in replies 👇 $NVDA $MU $GOOGL $MRVL $AVGO https://t.co/qyqtCnYDnz

J.P. Morgan: Recent investor discussion has been focused on delays in TPUv8t ramp, v9 tapeout and uncertainty on v10 allocation. Based on our checks, MediaTek's TPUv8t Zebrafish continues to ramp on schedule at TSMC, and remains on track…

What happens when you push a humanoid before it has learned to balance? In simulation it falls without damaging a single motor. Google DeepMind’s MuJoCo Menagerie includes a detailed Unitree G1 model with 29 degrees of freedom. Developers…

"Experiments in building products at Physical Intelligence" -- Quan Vuong's (@QuanVng) keynote at the Humanoid Hub Conference. He co-founded @physical_int, and before that helped build the foundational RT-2 and OpenVLA models at Google…

Microsoft Maia 300 is a hiring story as much as a chip launch. KORE1's take: the bottleneck isn't the silicon, it's the ML compiler engineers who make it useful. https://t.co/tFzlBZpkzh https://t.co/EDLcmTKqMP

Who designs the chip after Microsoft Maia 300? The same small pool Google, Amazon, Meta, and OpenAI are recruiting from. KORE1 breaks down which chip roles get squeezed first. https://t.co/tFzlBZpkzh https://t.co/liozv8z01k

Microsoft Maia 300 enters production in 2027, and the SIA projected about 67,000 U.S. chip jobs at risk of going unfilled by 2030. KORE1 expects compiler and validation engineers to feel it first. https://t.co/tFzlBZpkzh…

Citi: CSPs data center capex is expected to increase to ~$1.6 trillion in 2030E, with global capex spend increasing to ~$3.8 trillion. Google to see the strongest capex growth through 2030. $GOOGL $AMZN $META $MSFT

Microsoft just told the market it's betting multiple gigawatts of future data center capacity on chips it designed itself. Maia 300 claims 40 percent more tokens per watt than leading GPU based systems, and Maia 200, already running across…