17 Comments
User's avatar
Paul Parker's avatar

So we've just come to some fairly dramatic conclusions here, correct?

Does anyone want to offer an opposing perspective? Does anyone see any significant holes in his arguments or in his thesis? (Many thinks to dyb5785 for his framework- and AI-based analyis and elucidation.)

Expand full comment
Devansh's avatar

Part 2 will also discuss counters to the thesis. Coming soon

Expand full comment
dyb5785's avatar

IYH FWIW synthesis of counterarguments against Devansh's claims, integrating verified data, technical realities, and strategic analysis from NVIDIA's financial results and industry dynamics. https://notes.henr.ee/strongest-counterarguments-to-devansh-s-hardware-disruption-thesis-8jpt92

Expand full comment
dyb5785's avatar

IYH kindly find further notes on Devansh’s hardware investment thesis

“What Devansh identifies isn't competition—it's architectural phase transition from monolithic to disaggregated compute substrates.

Traditional taxonomy: CPU → GPU → Better GPU

Operational reality: Monolithic → Specialized → Orchestrated (ASICs + CXL + Photonics)

The critical insight: These aren't three separate markets but a dependency chain where each component's optimization creates bottlenecks that demand the next layer's evolution. This isn't hardware iteration—it's systems-level reorganization of computation itself.”

https://notes.henr.ee/the-fracture-points-in-the-post-nvidia-hardware-revolution-nthz42

Expand full comment
Devansh's avatar

Thank you for this. Are you sharing it with anyione?

Expand full comment
dyb5785's avatar

IYH I shared it w you. Use as you see fit - maybe for your part 2, you have been v helpful w your post.

Expand full comment
Devansh's avatar

I saw. It was quite well done. I wanted to know if there was a story behind it

Expand full comment
dyb5785's avatar
7dEdited

IYH Indeed there is. I have a specialized (Alex) Karpian analysis framework I synthesized from some of his public talks. I used your text as an input and said: "Execute analysis on Devansh Devansh Hardware Investment Thesis for the Post-NVIDIA Era of ASICs, CXL, and Photonics, Part I" and had two or three of them debate.

FWIW here's a provenance log (not all steps shows - claims needed to be per round fact-checked, responses weighed etc but the gist is here) https://notes.henr.ee/provenance-audit-trail-ai-infrastructure-investment-analysis-workflow-qgesig

Expand full comment
Devansh's avatar

You share that with Karp. I think he’d really like that you created a frmaework from his work

Expand full comment
dyb5785's avatar

IYH too kind. I don’t know the guy :D

TBH I am sometimes personally blown away how insightful his framework is. See eg this LI post (and the Google Wallet ZKP discussion w one of the MIT authors answering in the comments and the 1999 Hyperscalar paper verdict). https://www.linkedin.com/feed/update/urn:li:activity:7338656582714470400

Expand full comment
William Meller's avatar

I loved the depth you brought to this topic. But reading it, I kept wondering... It seems like the big players will have no choice but to jump early or risk being locked out of viable supply chains. But how does a midsize company even plan for that?

Expand full comment
Devansh's avatar

There's a lot of moving parts in the value chain. The midsize firms can quickly establish themselves to serve specialized cases there.

Expand full comment