SemiAnalysis

newsletter.semianalysis.com

Bridging the gap between the world's most important industry, semiconductors, and business.

Articles51

Ultra-High Interactivity on NVIDIA GPUs? - TileRT InferenceX

8/10/2026

Can TileRT software on NVIDIA GPU compete with Cerebras, Groq LPU, SambaNova? Batch Size 1, Disaggregated engine, high throughput engine Prefill, high interactivity engine decode

SpaceX 10GW in 2027 – Why It’s Real, Will Drive $300B ARR for SpaceX, and Why Microsoft Will Be the Largest Offtaker

8/7/2026

Inference at 100B/GW/year, SpaceX's stellar pace, Microsoft's 10GW 2026 Awakening, Azure Can Grow Tiple-Digits

Gemini is Cooked but GCP is Cooking

8/7/2026

or why DeepMind's long term failure is GCP's short term gain

Kimi K3, The Manos, The Mythos, The Legendos

8/3/2026

Kimi K3’s architecture: compressed memory, attention across depth, latent expert routing, and the inference performance

The Wild Wild West Of LEGO Datacenters

7/29/2026

The Labor Problem and Modularization to the Rescue

Can AMD break the CUDA Moat? AMD Advancing AI 2026

7/25/2026

Agentic Kernel Generation, Improvement in Software Quality, Unstable Internal Development Clusters, Helios MI455X Production Ramp Hell, Up to 105% Discounts from Finance Engineering

Vera Rubin NVL72 vs GB200 NVL72? Inference TCO & Architecture Analysis

7/23/2026

3 bit Rubin LUT Based Tensor Core, SM140 Feynman, Rack Scale, Perf Per MegaWatt, Perf Per Dollar, Software Improvements, Public Rubin Software, PyTorch, vLLM, OpenAI Triton

Meta’s Infrastructure Team Needs A Culture Reset

7/22/2026

Meta Infrastructure has become bloated, with middle managers expending resources on over-engineered technology solutions that lose sight of broader organizational needs.

The Future of Meta Superintelligence: A 1 Year Progress Update

7/9/2026

A top tier RL environment startup spawns out of thin air, the most aggressive compute ramp we've ever seen, 2000km+ scale-across, and some advice for Google DeepMind

Anthropic 3Q26 Profit Over $1B: The Anthropic IPO Financials Sneak Peak

7/8/2026

Anthropic’s Opportunity is Theirs to Lose

Nvidia GPU Debt Backstop Unleashes the AI Project Trinity: Capital, Offtake and Datacenters

7/6/2026

Over 7T AI debt by 2029, There can be no Neoclouds without the Trinity. Nvidia's Backstop Economics Explained. AI Debt Needs Quantified. Nvidia's Objective is to Broaden Compute Access, Develop AI Financing, Grow Neoclouds.

Meta Compute: Everyone Wants To Be A Neocloud

7/2/2026

Zuck Takes Plan B? SpaceX 2.0, Bedrock 2.0, MSL Isn't Giving Up, Scaling RecSys by 10x... ClusterMAX ranking coming soon?

EMIB-T Roadmap, Custom HBM, HBM4 Packaging Challenges, Microfluidic Cooling, Photonic Interconnects, and More

7/2/2026

ECTC 2026 Roundup, Intel, TSMC, SK Hynix, Samsung, Micron, Marvell, Lightmatter, Microsoft

TokenBudgeting: Our Conversations with Enterprises on Token Spend

6/30/2026

Was Widespread TokenMaxxing Ever Really Here?

US Grid Constraints: Towards 40GW+ of Behind-The-Meter Datacenter by 2028?

6/25/2026

Why the Grid Can't Keep Up, and Why that Drives Behind-The-Meter 50%+ of DCs/Year By 2028

China’s CXMT Is Set to Challenge DRAM Incumbents

6/23/2026

CXMT IPO, SKHynix, Micron, Samsung Competition, Process Node Deficit, China HBM, Wafer Adds, Memory LTAs

Stop Saying Half of 2026 US Datacenter Capacity Is Canceled

6/18/2026

Don't Believe The Vibecoded Estimates, Go Through Every Individual Filing Instead

RL Systems Mind the Gap: Matching Trainer and Generator Throughput

6/16/2026

RL Training Infrastructure, GRPO, PipelineRL, Async RL, Policy Staleness, RL Sandbox Infra, CPU Requirements, TCO Analysis, Thinking Machines Tinker

Is SMIC N+3’s Metal Pitch Smaller than Intel 18A’s?

6/14/2026

SMIC N+3 Node Deep Dive vs TSMC N6, TechInsights Private Equity Sale, SemiAnalysis Teardown Engineering & Evaluation Lab, HiSilicon Kirin 9030, Process Technology, Pattering, Cell Architecture

Intel Should Raise Capital

6/11/2026

Intel's woes are behind them. The heavy spending is ahead of them. Why an equity issuance in a hot equity market could make Intel so much better sooner.

DeepSeekV4 1.6T Day 0 to Day 43 Performance Over Time - Huawei, GB300 NVL72, MI355X, B200

6/9/2026

Day 0 Inference Performance, InferenceX, 100x performance improvement in 26 Days, Cost per Million Tokens, Huawei 950DT Inference Trace Analysis

China's Unitree Will Dominate Global Robotics

6/8/2026

The Fastest Iteration Cycle In Next-Gen Robotics Should See Unprecedented Acceleration

To Boldly Go: The Case for Space Datacenters

6/3/2026

Orbital Compute Model Launch

AI Dark Output: The Visible Cost of Invisible Output

5/29/2026

Why AI's increasing output is going to be one of the hardest economic measurement problems in history. AI "Dark Output" could end up being the majority of economic activity, but a challenge to measure

Finding Miscompiles for Fun, Not Profit

5/28/2026

Or: You don’t need access to Claude Mythos to spend $10,000 in an afternoon.

Anthropic Growth and Bedrock Mix Drive AWS Margins Higher While Peers Lag

5/27/2026

Amazon’s Bedrock Mix and Anthropic Deal Terms Combine to Show Greater Operating Leverage

Inside the 800VDC Revolution – Part 1

5/26/2026

Four-Phase 800VDC Transition, Power Rack Economics, SST, Equipment Content/MW Build, Supplier Implications

EDA Market Primer - Market Dynamics, Cadence, Synopsys, Siemens, China EDA Rise

5/21/2026

EDA Market size, Share, Business Models, Drivers, Changing Customer Base, Competitive Dynamics Across Synopsys, Cadence, and Siemens, China EDA, IP, Hardware, CoT, Lock-In Economics, Disruptive Forces

Cerebras — Faster Tokens Please

5/13/2026

// OpenAI and AWS Partnerships, Tokenomics Explainer, Architecture Deep Dive, Datacenter Ramp, Technical Roadmap

The EDA Primer: From RTL to Silicon

5/12/2026

Laying the Groundwork of the Current Chip Design Paradigm

AI Value Capture - The Shift To Model Labs

5/1/2026

A day in AI now feels like a year in any other industry. TSMC, Vera Rubin VR NVL72: V for Value

The Coding Assistant Breakdown: More Tokens Please

4/24/2026

Hands On With GPT 5.5, Opus 4.7, DeepSeek V4, Why Benchmarks Are Bad, and Who’s Going To Win

How Much Do GPU Clusters Really Cost?

4/20/2026

Calculating Cluster TCO, The Real Impact of Downtime, The Grand Unifying Theory Of Goodput, and a ClusterMAX 2.1 Update

ISSCC 2026: NVIDIA & Broadcom CPO, HBM4 & LPDDR6, TSMC Active LSI, Logic-Based SRAM, UCIe-S and More

4/15/2026

ISSCC 2026 Round-Up

The Great GPU Shortage – Rental Capacity – Launching our H100 1 Year Rental Price Index

4/2/2026

GPU Rental Pricing Dashboard Launch

Dissecting Nvidia Blackwell - Tensor Cores, PTX Instructions, SASS, Floorsweep, Yield

3/31/2026

Microbenchmarking, tcgen05, 2SM MMA, UMMA, TMA, LDGSTS, UBLKCP, Speed of Light, Distributed Shared Memory, GPC Floorsweeps, SM Yield

Nvidia – The Inference Kingdom Expands

3/24/2026

Groq LP30, LPX Rack, Attention FFN Disaggregation, Oberon & Kyber Updates, Nvidia's CPO Roadmap, Vera ETL256, CMX & STX

The Great AI Silicon Shortage

3/12/2026

TSMC N3 Wafer Shortages, Memory Constraints, Datacenter Bottlenecks, Supply Chain Wars Winner

Are AI Datacenters Increasing Electric Bills for American Households?

3/3/2026

Power prices misconception, PJM's poor market design, Capacity prices 9.3x growth,

Vera Rubin – Extreme Co-Design: An Evolution from Grace Blackwell Oberon

2/25/2026

Vera, Rubin, NVLink 6 Switch, ConnectX-9, BlueField-4, Spectrum-6, Seamless Cableless Compute Tray Design, Power Rack, VR NVL72 TCO and BoM

InferenceX v2: NVIDIA Blackwell Vs AMD vs Hopper - Formerly InferenceMAX

2/16/2026

The Artist Known as InferenceMAX. GB300 NVL72, MI355X, B200, H100, Disaggregated Serving, Wide Expert Parallelism, Large Mixture of Experts, SGLang, vLLM, TRTLLM

CPUs are Back: The Datacenter CPU Landscape in 2026

2/9/2026

RL and Agent Usage, Context Memory Storage, DRAM Pricing Impacts, CPU Interconnect Evolution, AMD Venice, Verano, Florence, Intel Diamond Rapids, Coral Rapids, Arm Phoenix + Venom, Graviton 5, Axion

Memory Mania: How a Once-in-Four-Decades Shortage Is Fueling a Memory Boom

2/6/2026

Pricing doubles again, Supercycle is bigger, and could last longer than you think

Claude Code is the Inflection Point

2/5/2026

What It Is, How We Use It, Industry Repercussions, Microsoft's Dilemma, Why Anthropic Is Winning

From Tokens to Burgers: A Water Footprint Face-Off

1/15/2026

A different perspective on the datacenter water debate, forget tokens/watt or tokens/dollar, it's about tokens/burger, how many In-N-Outs is equivalent to the world's largest datacenter

Interconnects Beyond Copper, 1,000 CFETs, SK Hynix Next-Gen NAND, 2D Materials, and More

1/13/2026

IEDM 2025 Round-Up

Apple-TSMC: The Partnership That Built Modern Semiconductors

1/8/2026

Wafer demand model, node economics, and the shifting power dynamics as AI reshapes the foundry landscape

RL Environments and RL for Science: Data Foundries and Multi-Agent Architectures

1/6/2026

Worker Automation, RL as a Service, Anthropic's next big bet, GDPval and Utility Evals, Computer Use Agents, LLMs in Biology, Mid-Training, Lab Procurement Patterns, Platform Politics and Access

Co-Packaged Optics (CPO) Book – Scaling with Light for the Next Wave of Interconnect

1/1/2026

Co-Packaged Optics (CPO) has long promised to transform datacenter connectivity, but it has taken a long time for the technology to come to market, with tangible deployment-ready products only arriving in 2025.

How AI Labs Are Solving the Power Crisis: The Onsite Gas Deep Dive

12/30/2025

Bring Your Own Generation, Sayonara Electric Grid, Turbines vs. Recips. vs. Fuel Cells, Why Not Build More CCGTs?, Onsite Power TCO

AWS Trainium3 Deep Dive | A Potential Challenger Approaching

12/4/2025

NL72x2/NL32x2 Scale Up Rack Architecture, Step-Function Software & System Improvements, Optimized Perf per TCO, “Amazon Basics” GB200 NVL36x2, Astera Labs, Trainium4