Austin Lyons (@austinsemis) / X

Austin Lyons

4,229 posts

Austin Lyons

@austinsemis

Substack: chipstrat.com | Podcast: semidoped.fm | Semis Analyst @creativestrat

chipstrat.com

Joined September 2007

Austin Lyons
@austinsemis
Sep 29, 2025
Insane precision from $ASML in this video. Watch from ~1:30 onward "Fun fact: the movement of the reticle stage and wafer stage are synced, but *light speed has to be taken into account*. The reticle stage needs to move slightly earlier, since it can shift more than 100
00:00
77K
Austin Lyons
@austinsemis
Oct 14, 2025
$META likes big racks. With $AMD MI450 inside.
29K
Austin Lyons
@austinsemis
Oct 20, 2025
“A lot of attention is given to compute, memory and networking in an AI data center. What gets less attention is the design of high capacity storage for AI workloads.”
23K
Austin Lyons
@austinsemis
Sep 18, 2025
$INTC Fundraising Tour (LBT's Version): 8/18 - SoftBank ($2B @ $23) 8/22 - US Gov't ($5.7B @ $20.47) 9/18 - Nvidia ($5B @ $23.28) Featuring special guests ("my good friends") Masa Son, Howard Lutnick, Jensen Huang. Where does the tour bus pull up next? $NVDA
35K
Austin Lyons
@austinsemis
Nov 11, 2025
$AMD sees >$1T TAM in data center by 2030
30K
Austin Lyons
@austinsemis
Oct 21, 2025
$CRDO example setup for $NVDA Rubin NVL144. Lots of opportunity with these high-density racks. 🧵
36K
Austin Lyons
@austinsemis
Aug 25, 2025
At @hotchipsorg, someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said “Yes...this was built on 18A. Everything we’re talking about for Clearwater Forest is on 18A... It’s real, or I wouldn’t be here..."
34K
Austin Lyons
@austinsemis
Sep 10, 2025
Hock also believes 1) Networking is the bottleneck at 1M+ clusters 2) Networking is harder than designing XPUs/GPUs 3) Broadcom is better at networking (presumably, better than $NVDA Spectrum-X, Infiniband, NVLink) $AVGO
Austin Lyons
@austinsemis
Sep 10, 2025
Hock Tan splits AI into just two $100B+ markets: 1) Foundation model labs 2) Everyone else $AVGO Broadcom is betting on #1 only. Tan gives a nice explanation of why merchant GPUs are great for #2
44K
Austin Lyons
@austinsemis
Sep 28, 2025
Toured the $INTC fab in AZ. 18A is real. Saw big machines. Wore the bunny suit. Lots of fun. 🇺🇸 Thanks @intel
21K
Austin Lyons
@austinsemis
Jul 18, 2025
Check out this footage of ASML's 2nm EUV tool in the Rapidus 2nm fab:
00:00
Austin Lyons
@austinsemis
Jul 18, 2025
Rapidus Achieves Significant Milestone at its State-of-the-Art Foundry with Prototyping of Leading-Edge 2nm GAA Transistors: First prototypes successfully demonstrate electrical characteristics as the company continues to hit targets leading up to 2027 mass-production TOKYO,
36K
Austin Lyons
@austinsemis
Aug 12, 2025
WOW, update your financial models!!! $CRWV implies a 7-year useful life for GPUs. H100s & A100s still earn $ after training contracts end, repurposed for inference. To be fair, Jensen has been saying this all along :) $CRWV $NVDA @CoreWeave @nvidia
27K
Austin Lyons
@austinsemis
Nov 4, 2025
$AMD CEO says CPU TAM increasing due to Gen AI. "Many customers are now planning substantially larger CPU build outs over the coming quarters to support increased demands from AI, serving as a powerful new catalyst for our server business."
21K
Austin Lyons
@austinsemis
Aug 28, 2025
$MRVL down 12% after hours. Earnings call was dominated by tough questions on the custom ASIC business. Here are some of the Q&As, styled by AI. A bit dramatic in the styling... but you’ll get the gist.
27K
Austin Lyons
@austinsemis
Aug 27, 2025
My takeaways from $NVDA earnings: • Networking = huge bright spot • China = upside optionality • Thor = underappreciated future play • Nearly $50B in a single quarter 🤯 • Air-cooled RTX Pro servers = nice incremental add • End-to-end AI system = Nvidia’s unbeatable edge
12K