• Semiconductors are Shkreli’s clearest actionable stance: he is long the sector, with NVIDIA (NVDA) positioned to benefit from rising compute demand, though AI efficiency gains could reduce hardware needs.
• Meta Platforms (META) is a qualitative AI investment idea, supported by its distribution and product development; no price target or timeframe was provided.
• Treat quantum-computing stocks and early-stage photonic startups as speculative, long-horizon themes—not near-term investments; Shkreli’s quantum shorts have lost millions, underscoring the risk of timing bearish bets.
• Private frontier AI companies and drug-licensing opportunities were discussed positively, but no accessible security, current deal, or valuation was identified.

Quantum Computing (Public Companies Not Specified)

• Shkreli is bearish on quantum-computing companies and said he had shorted them, though he also disclosed that the trades had lost millions of dollars.
• He argued that current quantum computers are slow and limited for most business uses, with a potential advantage concentrated in a narrow set of problems such as integer factorization and elliptic-curve cryptography.
• He criticized Google, IBM, and Microsoft for spending hundreds of millions of dollars annually on quantum computing without a clear near-term business payoff. He compared the risk to earlier computing technologies that became obsolete.
• He acknowledged the technology is early, but said the uncertainty around when useful commercial applications might emerge makes it difficult to time a short position.

Takeaways

• The discussion presents quantum computing as a high-risk, long-horizon theme, not a near-term commercial opportunity, in Shkreli’s view.
• His own reported losses underline the risk of shorting a speculative sector: even if the long-term thesis is bearish, timing can be difficult.
• No specific quantum-company tickers or price targets were given.

NVIDIA (NVDA) and Semiconductors

• Shkreli said he is long semiconductors and believes NVIDIA may continue to benefit as computing demand grows.
• He also cautioned that more-efficient AI models or other computing breakthroughs could reduce the amount of hardware needed for a given task, which could be negative for NVIDIA.
• He suggested that photonic computing could eventually outperform today’s conventional chips, while emphasizing that important challenges—including storage and nonlinearity—remain.
• He said current NVIDIA hardware generations could eventually become obsolete, but did not give a timeline or say that this was an immediate threat.

Takeaways

• The transcript supports a mixed but currently bullish view of semiconductors: Shkreli is long the sector, while recognizing that efficiency gains and alternative computing methods could challenge existing suppliers.
• Treat the prospect of photonic competition as a long-term uncertainty rather than a specific near-term catalyst; no price targets or timing were provided.

Photonic Computing Startups

• Shkreli highlighted NeuroFoss, OptiCore, and Lumai as startups working on photonic matrix-multiplication hardware.
• He said matrix multiplication performed using light could be extremely fast, and he and his wife are investigating the area, possibly including starting a company.
• He noted that photonic computing still has engineering challenges, including storage and nonlinearity.

Takeaways

• Photonic computing is presented as a potentially important alternative-computing theme, but the companies mentioned are early-stage and the transcript provides no commercial milestones, valuations, or investment recommendations.
• The stated technical hurdles make this a speculative area rather than an established substitute for conventional chips.

Meta Platforms (META)

• Shkreli was bullish on Meta, citing its distribution, product capabilities, and work on Muse.
• He viewed Meta’s product work as evidence that the company could find valuable ways to deliver AI to users.
• He described AI as being in its early stages and said the current chatbot experience may eventually look primitive.

Takeaways

• The positive case discussed for Meta centers on its ability to turn AI models into widely distributed products.
• The transcript does not provide a valuation, price target, or specific timeline, so the investment case remains a qualitative view rather than a quantified recommendation.

OpenAI and Anthropic (Private Companies)

• Asked whether he would buy OpenAI and Anthropic, Shkreli said yes. He said he already has exposure to OpenAI “through a few different ways.”
• He praised OpenAI’s strategy of pairing its models with hardware, arguing that delivering the model could be as valuable as developing it.
• He also described the potential for AI companies to build businesses across industries, while noting that the eventual value depends on whether more computing produces meaningfully better results.
• He argued that demand for intelligence and work could be very large, but the discussion also raised questions about AI valuations and how long providers might earn outsized returns.

Takeaways

• Shkreli’s view is bullish on frontier AI companies, particularly those that can combine strong models with effective distribution or hardware.
• For investors, the key uncertainties raised in the conversation are whether greater AI capability translates into lasting business value and how long current providers can maintain strong economics.
• These are private-company investments; the transcript gives no accessible security, valuation, or price target.

Biotech Drug Development and Licensing

• Shkreli described buying rights to drug candidates from larger pharmaceutical companies and developing them for different diseases.
• He gave sparsentan as an example: Bristol Myers had developed it for hypertension, and his team redirected it toward FSGS, a kidney disease that had not previously had an approved drug. He said the drug later received FDA approval and called it a blockbuster.
• He also described drug-asset transactions as an inefficient market where a small company may be able to identify value that a large pharmaceutical company does not prioritize.
• In a separate example, he said he bought a drug from Bristol Myers for $2 million and that the resulting company later became public with a value of $6 billion.

Takeaways

• The opportunity discussed is drug repurposing and licensing, especially when a candidate may have value in a disease area that its original owner did not pursue.
• The examples are historical anecdotes, not evidence that similar deals will succeed. The transcript does not discuss clinical-trial failure rates or provide a current drug-company recommendation.
• Shkreli was skeptical that AI will materially help drug discovery in the next 10 years, based on his experience building and analyzing drugs.

Royalty Pharma (RPRX) and Drug Royalties

• Shkreli described Royalty Pharma as a different kind of pharmaceutical investor: it buys rights to future drug cash flows at a discount.
• He cited its royalty on Remicade, saying the royalty was bought for roughly one times earnings and later increased in value by about 20 times.

Takeaways

• The conversation highlights drug royalties as an investment model that seeks returns from existing or developing medicines without relying solely on owning and operating a conventional drug company.
• The Remicade example is a historical success, not a forecast; no current valuation or recommendation for Royalty Pharma was provided.

Bristol Myers Squibb (BMY) and Pfizer (PFE)

• Shkreli used large pharmaceutical companies as examples of firms that may sell or license drug assets they do not prioritize.
• He said Bristol Myers sold him rights to a drug candidate for $2 million, and spoke generally about Pfizer selling drug rights for relatively small amounts compared with their potential value.
• These comments concerned the drug-licensing market rather than a view on either company’s overall stock.

Takeaways

• The investment theme is the potential value in underused pharmaceutical assets, particularly for smaller companies able to develop a candidate for a different use.
• The transcript does not give a bullish or bearish stock recommendation for Bristol Myers or Pfizer, or discuss current company valuations.

Artificial Intelligence Infrastructure and Compute Demand

• Shkreli argued that AI could create substantial demand for computing, including hypothetical uses such as a large company spending $10 billion on GPUs to improve business decisions.
• He also noted that greater efficiency could reduce hardware needs, creating uncertainty about how much of the expected demand will accrue to infrastructure providers.
• He compared some of the broader benefits of AI to the internet, suggesting consumers and businesses may capture much of the value even if infrastructure suppliers do not.

Takeaways

• The transcript presents AI as a potentially large growth theme, but distinguishes strong demand for AI services from certainty about which infrastructure companies will capture lasting profits.
• Investors should consider both sides raised in the discussion: expanding use cases could support compute demand, while more-efficient models could reduce the hardware required per task.