Intel’s EMIB vs. TSMC’s CoWoS, prefill and decode disaggregation, 576 GPUs in a rack, edge compute for end-to-end transformers, AI-native EDA
Independent semiconductor research
Chipstrat is independent research on semiconductors and AI infrastructure.
I’m Austin Lyons. I write the Chipstrat newsletter, co-host the Semi Doped podcast and work with semiconductor companies as an independent analyst. I trained as an electrical engineer, worked at Intel, built software startups and later earned an MBA. I cover how the technology works and why the business decisions around it get made.
- MSEE, University of Illinois Urbana-Champaign
- MBA, University of Iowa
- Formerly at Intel
- 13,000+ subscribers

Latest
- Is EDA Dead?
GPT-Synopsys and whether model labs can replace EDA tools
- AI With No Decode (Jev), and What It Means for AI Infrastructure
Jev from TypeSafe answers in one pass instead of writing text token by token. Why LLMs are hard to use inside software, how Jev fixes it, and what a model with no decode phase means for AI Infra
- 🎙️The Race to Build the Next Trillion-Dollar AI Chip Company
TechSurge: Deep Tech VC Podcast
- A Billion Agents Need Somewhere to Run
Grok Bot, Meta Muse, and what consumer adoption means for CPUs, accelerators, memory, and advanced packaging.
- Oligopoly Equilibrium
Why most semiconductor markets settle at two to three players
Research
164 posts since 2023
GPUs, XPUs, CPUs
Merchant GPUs, custom accelerators and the CPUs that sit beside them. Nvidia, AMD, Broadcom, Marvell and the AI chip startups.
61 posts →Optics & Networking
How AI clusters are connected. Scale-up and scale-out networking, SerDes, AECs, pluggables, LPO and co-packaged optics.
19 posts →Foundry
Leading-edge manufacturing and who controls it. TSMC, Intel Foundry, Samsung, process nodes and lithography economics.
26 posts →Memory
DRAM, HBM and NAND in the AI buildout, and what memory supply means for system design and pricing.
5 posts →Chiplets & Packaging
Advanced packaging and the chiplet ecosystem. CoWoS, EMIB, die-to-die interconnect and the economics of disaggregated design.
8 posts →Autonomy / Physical AI
Silicon and models for vehicles and robots. Rivian, Qualcomm automotive, LiDAR and vision transformers.
10 posts →Edge AI & Devices
Inference on phones, PCs and embedded hardware. NPUs, tokens per second per watt, Apple, Qualcomm and Arm.
21 posts →Big Tech & Cloud
Hyperscaler capex, in-house silicon and AI demand. Microsoft, Amazon, Google, Meta and OpenAI.
14 posts →Fundamentals
Primers that assume no engineering degree. How chips, networks and optics work, and why the trade-offs matter.
22 posts →Interviews
Conversations with the engineers and executives building this hardware.
13 posts →Most covered companies
Readers
Who reads Chipstrat
Industry
- Hyperscalers, neoclouds and AI labs
- Chip companies
- TSMC
- AI chip and photonics startups
- Etched
- Lightmatter
- Substrate
- Equipment, EDA and materials

- Lam Research

- Systems, networking and OEMs
Investors and research
- Buy side
- D. E. Shaw
- Tiger Global
- Sell side
- Venture capital

- Thrive


- Consulting, research and media
- SemiAnalysis
- The Information
What readers say
“I supported Chipstrat as I find the articles interesting, and our CTO and Chief Strategy Officer recommended it.”
“I really value Austin’s work! Indispensable.”
“Thoughtful writing on all things chips.”
“Austin’s clarity of thinking is exceptional, and his writing style is crisp and concise.”
Analyst
I work with semiconductor companies as an independent analyst.
Companies retain me for advisory calls, briefings, executive sessions and events. They get a candid outside view from someone who talks to their customers, competitors and investors every week. Coverage in Chipstrat is never part of a contract.
Companies I work with or have worked with as an analyst
Disclosure
Chipstrat is published for information and analysis. It is not investment advice. I have paid relationships with some of the companies I cover, and I list them on this site. No company pays for coverage.




