Google Cloud Pushes Agentic Cloud to Outpace AWS, Microsoft
.@GoogleCloud Next opens Wednesday in Las Vegas with an opening keynote titled “The agentic cloud.” Google Cloud exited Q4 2025 with the fastest growth of the Big Three. Now Thomas Kurian needs to prove Google can turn TPU and model momentum into a durable enterprise agent-platform share against @awscloud Bedrock AgentCore and @Microsoft Foundry Agent Service. Ten things I’m watching at Google Cloud Next: 1/ Next Gen TPU details if anything new: claimed perf, process node, memory profile, OCS topology, and shipping window. 2/ @AnthropicAI on stage with a real 2026 deployment milestone, not just a restatement of long-range TPU demand. 3/ Newly named Fortune 500 Gemini Enterprise wins with 12-month steady-state ROI, deployment scale, and cycle-time improvements beyond the Mercari and Swarovski examples. 4/Google’s agent control-plane depth versus Amazon Web Services (AWS) Bedrock AgentCore and Microsoft Foundry Agent Service, especially policy, evaluations, memory, identity, and orchestration. 4/ Axion’s roadmap beyond C4A and N4A, plus credible first-party Axion-versus-Graviton benchmarking. 5/ Agent2Agent shown live with real cross-stack handoffs across major enterprise platforms beyond ecosystem logos and partnership slides. 6/ TPU availability inside Google Distributed Cloud and net-new sovereign customer wins beyond the roster Google has already talked about. 7/ @Intel and Google moving from partnership language to shipping product: Xeon 6 instance GA, IPU roadmap, and what actually becomes available now. 8/ Gemini Enterprise pricing that reflects agent consumption or metered usage, not just a repackaging of existing Starter, Business, Standard, Plus, and Frontline tiers. 9/ How TK talks about NVIDIA with TPU deals cut everywhere. Google does not need to prove it can build infrastructure. It needs to prove it can own the enterprise control plane for agents, show measurable customer value, and make the monetization model easy to understand.
Samsung Strike Threat Boosts Micron Stock Outlook
Positive call out on $MU Samsung / MU: Quick thoughts on Samsung worker unrest What happened: According to credible reports out of S. Korea, unionized workers of Samsung Electronics are planning to rally outside its manufacturing facilities due to breakdown in wage negotiations....
TSMC Controls AI Chip Supply; Packaging Limits Scale
AI Infrastructure Stack (and who’s actually capturing value) Let’s break down the stack 👇 Semiconductor Foundation (Foundries & Equipment) $TSM TSMC Fabricates nearly all advanced AI accelerators globally. 3nm capacity is heavily reserved for AI workloads, while 2nm ramps require new GAA...
AI Shifts From Coding to Autonomous Hardware Design
Last year: AI started writing code. This year: AI is starting to design hardware. And not just visually correct designs functionally correct systems with motion and constraints. We’re watching CAD evolve into an autonomous layer. The engineers who thrive will: Think in systems Define...
AMD Poised to Ride AI Chip Boom to $600
Bullish take on AMD: a deep-dive analysis argues the AI chip market is set to explode, and AMD's tech could drive its stock to $600. A bold, long-term vision. Investing
Google Partners with Marvell for Next‑gen AI Inference Chips
Google in Talks With Marvell to Build New AI Chips for Inference — The Information https://t.co/yPq06Iptig
Tiny BCI Chip Delivers High‑Bandwidth Wireless Neural Links
Ultra-Compact Brain–Computer Interface Chip Enables High-Bandwidth Wireless Neural Communication by @CUSEAS #MedTech #Healthcare #HealthTech #Tech #Technology https://t.co/ogVf9xUrki
Intel's RISCified X86-64 Beats AI Accelerators
Intel won by RISCifying their CISC - still amazed how well x86-64 worked out LPUs is a subset of AI, an accelerator, it's not the RISC @tenstorrent , Trainium, Google TPU are closer. Clean Tensor processor is step one. Then generality, memory...
Semiconductor Technical Roadmaps Guide Future Manufacturing
#Technology #Newsletter #Semiconductor #Manufacturing #Roadmaps NLOG-300 | Semiconductor And Beyond Newsletter | The Semiconductor Technical Roadmaps: https://newsletter.chetanpatil.in/p/semiconductor-and-beyond-newsletter-300/
AVGO and MRVL Command 80% of Custom Chip Market
$GOOGL $MRVL - good news but not surprising as $AVGO and $MRVL have an 80% share of the custom chip mkt. https://t.co/rk3fJlTbtA
Beyond Moore's Law: Shaping Tech's Next Evolution
Moore’s Law has defined the pace of development in tech for over half a century - where do we go next? https://t.co/sJM3bYtWxZ
Data Platforms Transform Semiconductor Manufacturing Efficiency
#Technology #Blog #Semiconductor #Manufacturing #Data BLOG-323 | The Emergence Of Data Platforms In Semiconductor Manufacturing: https://www.chetanpatil.in/the-emergence-of-data-platforms-in-semiconductor-manufacturing/
TT‑Lang: Python DSL for Tenstorrent’s High‑performance Kernels
TT-Lang from Tenstorrent from Groq? It’s a Python-based DSL that lets you write high-performance custom kernels and fused ops directly on Tensix cores, Blackhole, etc.. Think “Triton but made for Tenstorrent hardware.” V 1.0 next week
US Growth Rides on Fragile TSMC Dependence
With robust resilient global supply chains, American growth has seen record gains, including with 90 percent dependence on TSMC for advanced node production. No one explains how much manufacturing will strengthen the economy and at what cost, economics matter....
Intel's Packaging Wins Set Stage for Wafer Deals
Intel has made great strides. I was one of the only analysts that gave them a shot when the company had been written off for dead by most. They’re definitively getting advanced packaging from big XPU players. I called this...
US Blocks Chinese LED Chipmaker’s Dutch Acquisition
Chinese LED chipmaker’s purchase of Dutch firm collapses after US opposition The specific national security concern not clear. What about " small yard, high fence"? LEDs? https://t.co/4JNTUcmAyP
Space‑based Datacenters Will Dwarf Earth GPUs, Boost Security
Tonight I hung out with an entrepreneur who is building https://t.co/1tXBvxF3GP Datacenters in Space. He is putting 42,000 up over next few years. All put in space by @SpaceX And all packets will go through @Starlink. GPU inside made by @nvidia. He said the Vera...
Europe’s AI Chip Push Gains Momentum, Still Funding‑Limited
Europe’s AI chip ambitions are gaining momentum, but face structural limits. Startups like Euclyd are seeking significant funding to scale, reflecting strong demand in the sector. The challenge is clear. Europe has the talent and intent, but funding depth and manufacturing capacity...
China's Limits: Why It Can't Replicate ASML
Why China Can't Copy ASML? https://t.co/A0c2vGrU8i via @YouTube #microprosessor #chip #ASML #tech #semiconductor @sonu_monika @enilev @Jagersbergknut @TysonLester @chidambara09 @labordeolivier @BetaMoroney @tlloydjones @Nicochan33 @jeancayeux @RLDI_Lamy @pierrepinna @pierrecappelli @pchamard @JeromeMONANGE @thierry_pires @MaiaGabunia @amalmerzouk @NewsNeus @mary_gambara @PawlowskiMario @XavierAncelin @drsharwood @kalydeoo @AnthonyRochand @Sharleneisenia @1OFFGINGER @smaksked...
Semiconductor Roadmaps Align Industry on Future Scaling
#Technology #Thread #Semiconductor #Manufacturing #Roadmaps The Semiconductor Let Us Explore Roadmaps: 1/ - Semiconductor Roadmaps Define The Long-Term Direction Of Technology, Manufacturing, And System Innovation. - They Align Industry Players On Scaling Challenges, Investment Priorities, And Ecosystem Evolution
Exynos 2600 Beats Snapdragon 8 Elite Gen 5 by 20fps
Exynos 2600 once again outperforms the Snapdragon 8 Elite Gen 5 in gaming Achieving 102fps, over 20fps more than the 8 Elite Gen 5, is an incredible result From @ luancastro_lc on X
NVIDIA Unveils Sub‑millisecond Always‑on Facial‑recognition Chip
@NVIDIA has a new chip that can detect human faces in less than a millisecond and can remain on at all times. https://spectrum.ieee.org/face-recognition-nvidia-chip-soc
SaaS Recovery and Nvidia Tied to Iran Volatility
SaaS-pocalypse' Recovery & NVDA Strength Hinges on Iran Volatility https://t.co/626qaLOTay @SchwabNetwork “All victims of the "SaaS-pocalypse" are being looked at once again” - @rwang0 @constellationr
Custom Chips Threaten True Openness of AI Models
If the most performant AI models are going to be those that are being run on a custom-tailored chip architecture (for that model) for inference, does that mean that "open source" means less than meets the eye, since most users...
AI Redefines TSMC’s Capacity and Global Strategy
In other words, TSMC is no longer dealing with a normal semiconductor cycle. It is dealing with a structural reallocation of capacity. AI is reshaping the logic of semiconductor capital spending, extending the lifecycle of leading-edge process nodes, and even changing...
Opus 4.7 Redefines CAD with Autonomous Complex Design
Opus 4.7 sets a new benchmark in agentic CAD by handling complex design workflows with minimal supervision
Geopolitics and AI Redefine Semiconductor Success Strategies
The semiconductor industry is no longer just about technology. Supply chains and geopolitics are now at the center of strategic decisions, with factors like export controls, geographic concentration, and material dependencies shaping which companies can scale and succeed. These risks create...
Intel's Revival Would Boost Biden, Trump, Industrial Policy
If Intel turns around it will be an incredible success story for Biden/Trump and industrial policy. https://t.co/8hNrsePZVO
RAM Crisis Fuels Fake SSDs and CPU Sales Slump
More fake SSDs and a CPU sales slump — the casualties of the RAM crisis are piling up, sadly. https://t.co/sdQQblnVyL
Compute Shortage Drives AMD, Amazon, Cerebras Deal Rush
Chatter being $AMD could get a deal with Anthropic which makes total sense given the compute shortage. I have been saying for a while I think Amazon is a good candidate also for Helios. Bottom line, all available viable compute will...
Anthropic May Add AMD MI450 GPUs to Stack
Rumour: Anthropic is planning to incorporate $AMD GPUs into their hardware stack for the first time starting from MI450.
GPU Is the New X86; TPU Dominates AI
The reality is that that the GPU is the new x86. It’s general purpose. It has to be able to do a lot of things fairly well, but it’s an expert at nothing. The TPU is purpose built to run...
Nvidia's AI Demand Shifts to Massive Deployment Wave
Nvidia's AI story is getting a major upgrade. Analysts are now convinced as demand shifts from just training models to the massive, long-term phase of deploying them everywhere. This "second wave" could be even bigger. AI
No Detuned GPUs Pushes China to Build XPUs Faster
Agree 100% with Gavin on this one. 1/ Yes China has its own aspirations. The question was how quickly they needed an alternative and the mandates and investment. 2/ NOT selling detuned GPU editions means they’re forced to do the...
China Doubles AI‑Science Compute Without US Chips
China doubles ‘AI for science’ computing scale in 2 months using no US chips | South China Morning Post https://t.co/3H0S5aW1UM
Intel Nears ATH; Terafab Deal May Redefine Foundry Landscape
Intel ~9 % away from ATH. I also think the terafab deal may be more consequential in many respects than is realized today. Wait until you see my foundry model :)
Chip Packaging: The Glue Powering Modern Electronics
Inside every laptop or desktop is a circuit board filled with tiny electronic components, many of which are silicon chips. These chips may look like simple rectangular blocks, but on their own, they cannot do much without being connected to...
AMD's Massive Growth Won’t Require Nvidia Share
$AMD doesn’t have to take share from $NVDA to achieve massive DC growth. This was always going to happen. 👏🏻
Limited Chip Exports Preserve US Edge, Fuel China’s Future AI
Jensen's arguments against chip export controls didn't make a ton of sense to me, but here's a steelman: selling a limited quantity of n-1 generation chips to China could undercut their domestic industry, while retaining a 10x American compute advantage. Totally...
TSMC's Historic Quarter: What Happened and What’s Next
Join the Culpium substack chat in 4mins. TSMC's historic quarter. What just happened, and what's to come. Subscribers only (but it's free) https://t.co/vqv9GQYyVZ
TSMC Signals New LPU Bid, Samsung's Groq Order Uncertain
TSMC hints at next-gen LPU bid, stokes speculation that Samsung's Groq order may not last https://t.co/hGNZ3E1yBn
Midwest Boutique Says DRAM Price Forecasts Stabilizing
Did Midwest Boutique send a cautionary commentary out on $MU ? "Our work suggests conventional DRAM price forecasts are stabilizing somewhat after continued meaningful upside"
TSMC Defies Slow Critiques, Shows Strong Earnings
Some say TSMC is too slow. But earnings today show it’s stronger than ever, even as a bold new entrant joins the chip race. CC Wei 🐢 @elonmusk 🐇 https://t.co/mrlSHR8taT https://t.co/FtqYRRTCxQ
China Can Overcome Chip Process Gap by Stacking More
Jensen Huang: The claim that China lacks AI chips is utterly baseless; they can scale up computing power infinitely if they choose to. China can fully compensate for the process gap in individual chips by stacking more chips. 'They have...
Compute Strategy: Build Your Own AI Infrastructure
Everyone's debating AI models. Almost no one asks what they run on. XPU co-development. Custom Ethernet for AI. Boston Dynamics in production. Compute is now strategy. Rent it: ceiling set by someone else. Build it: you write your own. https://t.co/U5DIK1nJt6
Semiconductors Shift Computation Closer to Memory
#Technology #Thread #Semiconductor #Manufacturing #Memory The Semiconductor PIM To PNM: 1/ - The Semiconductor Industry Is Rethinking Where Computation Happens. - Traditional Architectures Move Data Back And Forth Between Memory And Processors, Driving Latency, Power, And Cost.
Every AI Chip, Not Just NVDA or AVGO, Will Sell
It’s not $NVDA or $AVGO. It is every AI chip that can be built will be sold. 💪🏻🚀 https://t.co/K0jvKcT7SF
CEO Discusses Monetization and New Agentic EDA Stack
Here at #CadenceLIVE, CEO @OfficialADevgan held an Analyst Q&A session - we had questions about monetization but also about the agentic stack coming to EDA. Full transcript now available at More Than Moore. https://t.co/8poqXP70GG https://t.co/nj1YcE6nGL
Machine-to-Machine Communication Will Skyrocket Processing Demands
General-purpose processing demands will multiply once machines are talking to machines. https://t.co/J4yo86mxLl #CPUs @Arm #AgenticAI #chiplets #NPU #GPU #datamovement
TSMC Q1 Earnings Preview Amid Record AI Compute Surge
Good morning from Taipei. TSMC reports 1Q earnings & hosts an IR call later today. The AI compute sector is at its hottest level on record. I’ll have full coverage soon after the call. Straight to your inbox from https://t.co/yt0PGDxqqi