Gen AI delivers its greatest impact when applied strategically in operations helping drive digital transformation.
This collection of insights explores its adoption across the operations value chain, enhancing both employee and customer experiences. ⬇️ https://t.co/ALSTIi57wM
Sharing how I view the AI Semiconductor Value Chain:
Design
The chip designers like $NVDA and $AMD are the primary revenue beneficiaries.
Companies like $AVGO earn revenue from helping non-chip companies like $GOOGL to develop chips.
EDA and IP providers supply the tools necessary to design chips.
Manufacturing
The manufacturing ecosystem also gets a boost from AI, but it's to a lesser extent.
TSMC derives ~6% of revenue from AI, according to management.
The semicap and WFE ecosystem then provide tooling to the foundries.
Finally, chips are tested and packaged before being shipped.
The final chip used in training or inference depends on the specific AI workload.
The Semis layer makes ~90% of all Gen AI profits today!
Where does future value accrue and how do we get there?
I explore where the $$ in Gen AI is today (semis), where it's headed (apps), and how it compares to prior tech breakthroughs.
DeepSeek founder Liang Wenfeng:
>> Studies machine vision at Zhejiang University
>> At 30 in 2015, launches High-Flyer quant hedge fund
>> Makes a fortune (now $8B AUM)
>> Wants to build “human” level AI as side hustle and pitches partners but they initially sceptical
>> Buys 10,000 H800 chips in 2021 and brings over his top hedge fund employees (all have tons of experience squeezing juice out of Nvidia GPUs for the fund)
>> Launched DeepSeek in 2023 and hires dozens of PhDs from top Chinese universities (Peking, Tsinghua and Beihang)
>> Pays top top top salary for tech talent only matched by Bytedance in China…wants DeepSeek to be leading “local” company
>> US export restrictions force DeepSeek team to get creative and they do, finding new training methods to make LLM models (V3, r1) competitive with OpenAI, Anthropic, Gemini, Grok, LLama etc at ~1/20th the cost
>> Training costs not exactly apples-to-apples but novel methods and clear improvements in efficiency (also questions around copying other models, larger H-100 clusters they maybe can’t talk about and/or CCP support)
>> Open sources and publishes methods (r1 reasoning paper has 200+ authors)
>> DeepSeek just hit top of App Store
***
FT: https://t.co/C6Ry6QPvnv
China has created one of the world’s best AI models for only $6 million, as opposed to the billions spent by Facebook, Google, Microsoft etc.
And DeepSeek is open-sourced, while the US models are proprietary and secretive—exposing the West’s bloated, profit-driven approach to innovation.
> be an electric engineering student
> team up w/ cracked classmates
> start quant trading
*we’re so cracked*
> founded a quant firm in his 30’s
> makes ¥100B trading with ai/ml
*we’re even more cracked with ai*
> buys thousands of Nvdia GPUs
> creates DeepSeek as a side project