Martin Shkreli closes $5M seed round, pitches photonic computing as the next Nvidia challenger

Summary

Martin Shkreli has closed a $5M seed round for his financial software company, following an earlier $2M raise approximately 18 months prior. The company is near profitability and is targeting a Series A of $10M to $30M later in 2025. Shkreli frames financial software as an unusually forgiving market because traders are perpetually willing to trial new tools, and argues the sector suffers from a domain expertise gap that incumbents like Bloomberg have not closed.

A flagship product in development, internally referred to as a situation monitoring suite, would go significantly beyond Bloomberg's existing geopolitical tracking tools. Shkreli envisions hedge funds at the scale of Citadel or Millennium tasking private satellite constellations to gather proprietary intelligence, treating the constellation cost, estimated around $100M, as accretive to EBITDA against Bloomberg's $8B to $10B revenue base. He cites the Jeff Bezos satellite strategy as the conceptual origin of this direction.

Prediction Markets and Private Company Pricing

Shkreli is skeptical of current prediction market structures, comparing the sector's maturity to proto-social media before Friendster. He singles out Kalshi specifically as a competitor he does not wish well. His core critique is that binary outcome markets are economically irrational at scale, arguing that a continuous dollar-for-dollar market tied to private company valuations would be more useful. He has been running a market-making operation on Polymarket focused on Bitcoin 15-minute price intervals, which he describes as mostly a curiosity rather than a serious instrument.

On private company pricing, a public dispute with Joe Lonsdale over Ramp's valuation on Shkreli's platform produced a pointed observation: investors with deal access effectively hold insider information, and expecting a public-facing market to reflect that information in real time is structurally unreasonable. SpaceX prices on his platform tripled after Elon Musk announced a buyback at what the market perceived as a dramatically elevated price.

Photonic Computing as an Nvidia Challenger

Shkreli is actively researching photonic computing as a post-GPU compute thesis, describing it as a field with a 50-year history that has been consistently underinvested relative to its potential. The core technical argument is that matrix multiplications, which represent roughly 90 to 95% of what a GPU executes, can be performed by light through constructive and destructive interference at up to 100 terahertz, naturally and without the overhead of general-purpose instruction sets. He frames this as a more commercially viable near-term bet than quantum computing.

Companies he is monitoring include Lightelligence, a Singapore-based firm with MIT origins and partial China ties, and Light Matter, whose commitment to all-optical photonics he considers unclear. Marvell's acquisition of Celestial AI is cited as evidence that photonic interconnects are already migrating into mainstream chip infrastructure. He draws an explicit analogy to the Grok acquisition at $20B and quantum computing names trading at $10B to $20B valuations, arguing that any credible challenger to Nvidia's compute dominance could be worth trillions, making even low-probability bets rational for retail investors who cannot access private rounds in vehicles like Grok or Unconventional at its $5B raise.

AI in Drug Discovery

Shkreli's view on AI and pharma diverges sharply from consensus. He does not believe AI materially accelerates small molecule drug discovery or protein design, arguing the forcing functions already in place are sufficient. The more actionable opportunity, in his framing, is organizational. Large pharma operations typically run 10 to 15 or more siloed departments, including toxicology, pharmacokinetics, and preclinical teams, whose coordination failures drive irrational capital allocation. He estimates that up to half of all drugs currently in clinical trials should not have advanced that far, with the decision driven by competitive pressure and Wall Street optics rather than data.

Clinical trial execution, not headcount, represents the dominant cost driver once a compound is identified. He characterizes the pre-trial drug invention phase as costing $10M to $50M, with the following $400M to $500M consumed by trial execution. His comparison of Chinese versus US biotech operations is striking: a mid-sized American biotech runs 100 to 300 people while a comparable Chinese firm may employ 6,000, including 800 chemists, funded by equivalent capital raises of around $300M in Hong Kong markets. He attributes China's accelerating competitive position in drug development to this structural difference in human capital deployment and sees it as a genuine long-term threat.

Eli Lilly's recently announced AI partnership, valued at approximately $1B, is noted as a signal that the industry is beginning to take the operational AI thesis seriously, even if most external commentators focus on the wrong layer of the stack.