All-In Podcast · Oct 12, 2026 · 25 min
Martin Shkreli: Big Pharma's Blind Spot, Betting Against Quantum & What Prison Taught Him
Martin Shkreli argues that quantum computing is commercially overhyped: its advantage is narrow, current machines are too limited for broad workloads, and he has lost millions shorting quantum companies. He is bullish on AI’s long-run potential to expand intelligence and productivity, and says Meta and OpenAI are well positioned, though he warns that more efficient computing could pressure Nvidia and that today’s AI chatbots are only an early phase. He also argues that AI safety fears are exaggerated and doubts AI will materially advance drug discovery in the near term.
Where they stand
Shkreli is bullish on risk assets in the near term and dismissed a recent sell-off as a head fake, while acknowledging that technology winners and market timing are uncertain.
Medium conviction · weeks
He expects AI to create enormous productivity and business opportunities, while judging safety alarmism and near-term AI drug-discovery claims as exaggerated.
High conviction · years
Crypto prices and adoption are not discussed as an investment outlook, although Shkreli notes quantum computing could eventually threaten cryptographic problems.
Low conviction · years
Key takeaways
- Shkreli says quantum computers have a narrow useful advantage, are not general-purpose faster computers, and have yet to demonstrate even basic tasks he tried; he has shorted quantum companies and lost millions. 2:39
- He sees photonic computing as a possible future alternative to GPU-based matrix multiplication, potentially making today’s Nvidia hardware obsolete over a long horizon. 6:54
- He remains bullish on AI’s potential, arguing that demand for intelligence and work is effectively unbounded and that AI could spawn many businesses. 16:47
- He identifies Meta as a likely AI winner because of its distribution and product capabilities, and says he would invest in OpenAI as well. 18:56
- He says AI safety warnings are overblown and predicts AI will not produce a relevant new drug in the next 10 years. 15:09
- He says AI efficiency gains could be bad for Nvidia, even as he remains long semiconductor-related investments for now. 11:53
Calls and positions
- Shkreli has shorted quantum-computing companies, but the position has lost millions of dollars so far.
“we lost millions of dollars shorting quantum computing companies”
- Shkreli expects today’s Nvidia accelerators to be obsolete in about 20 years as better computing methods emerge.
“in 20 years these H100s are in the scrap heap”
- Shkreli says AI could create entire portfolios of businesses within companies such as OpenAI and Anthropic.
“you're going to have a whole S&P 500 within Open AI or Anthropic”
- He says AI will not produce a drug relevant to the field within the next 10 years.
“I don't think there'll be a single AI-relevant drug in the next 10 years.”
Views expressed
- Modern quantum computers are too slow and limited to outperform conventional computers on general-purpose tasks such as sorting.
“The modern quantum computers run at a clock speed of hundreds of hertz or thousands of hertz”
- Quantum computing’s notable algorithmic advantage is narrow and does not provide a broad commercial reason for most businesses to adopt it.
“unless you were in the business of factoring numbers, there's there's not a whole lot uh to be excited about here.”
- Photonic matrix-multiplication systems could eventually outperform Nvidia’s GPU approach, though Shkreli says important technical challenges remain.
“matmul's in light are free.”
- He says much more efficient transformers would be negative for Nvidia, although he is currently long semiconductor-related investments.
“If there's a thousand times more efficient transformer, that's bad for Nvidia.”
- Shkreli is bullish on the market and considered a recent sell-off to be a head fake.
“Yeah, I think it was ridiculous.”
- He argues that demand for intelligence and work has no practical limit, supporting substantial long-term AI opportunities.
“there's no limit towards intelligence.”
- He expects Meta to be a leading AI winner because it has distribution and has demonstrated product capability.
“I think it's meta.”
- Shkreli says he would invest in OpenAI and believes its combination of strong models and hardware delivery is strategically valuable.
“Yeah. Uh yes, so we have Open AI obviously”
- Shkreli argues that AI catastrophe warnings are premature and that AI is software rather than an imminent existential threat.
“this is just software.”
- Shkreli says quantum computers could perform integer factorization and elliptic-curve discrete-log calculations in polynomial time, a potential concern for encryption.
“the quantum computers can do those in polynomial time”
Topics covered
More from All-In Podcast
- AI Cyber Panic, Ken Griffin’s Bubble Warning, The Fall of France?Oct 23, 2026
- Arvind Krishna & Nikesh Arora: Quantum Threat, Open vs Frontier AI, AI Cyber Warfare & SaaS CollapseOct 21, 2026
- Dylan Field: Is AI Replacing Designers?, Design is the New Code & Democratizing CreativityOct 19, 2026
- CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags, Bill Gates DoomerOct 17, 2026
Summary generated by AI from the video's captionsand checked against it; quotes are verbatim. Processed Oct 25, 2026.
Read on Risk summarises what other people said in public videos and podcasts. Summaries are generated by AI and checked by code against the transcript, but can still be wrong; follow the source links before relying on anything. This is not investment advice and nothing here is a recommendation to buy or sell. Disclaimer ·How this works