Martin Shkreli's New Computing Firm Is Betting It Can Upend Nvidia's Business Model

“Pharma Bro” Martin Shkreli has published a detailed bull case for photonic computing Thursday , sayinghe is now actively working at Q/C Technologies Inc. (NASDAQ:QCLS), the micro-cap firm he first invested in last December.

What Shkreli Is Arguing

Shkreli’s Substack essay, titled “Photonic Computing: The Final AI Hardware Frontier,” lays out a simple thesis: GPUs from Nvidia Corp. (NASDAQ:NVDA) spend roughly 99% of their time doing one thing, matrix multiplication.

Trillions of dollars in market value exist to make that single calculation faster and cheaper.

Photonic computing uses light instead of electrical signals to do the same math.

Shkreli argues light can handle these operations in near-linear time versus cubic time for GPUs. On large matrices, that difference could be enormous.

Shkreli was named a strategic advisor to QCLS in December. His partner Chelsea Voss, a member of technical staff at OpenAI, joined the board in January.

Why It Matters For The GPU Trade

If Shkreli is right, some large prediction market contracts in crypto would need to reprice.

Polymarket gives Nvidia a 61% chance of ending 2026 as the world’s largest company by market cap, and the AI bubble burst contract sits at 19% by year-end , with one trigger being NVDA falling 50% from its all-time high.

Both markets assume GPU dominance holds.

Shkreli is betting it doesn’t. If photonic computing scales, GPU pricing power could compress.

NVDA is trading at roughly 25x forward earnings despite 67% EPS growth, a valuation that Bernstein’s Stacy Rasgon called historically cheap but one that still prices in GPU dominance continuing indefinitely.

Even if photonics eventually disrupts GPU economics, the prediction markets suggest almost nobody sees it as a near-term threat.

Why Traders Are Skeptical

QCLS started life as TNF Pharmaceuticals and pivoted to photonic computing in September 2025.

SEC filings explicitly earmarked it for “cryptocurrency infrastructure.” Repackaging a crypto pivot as an AI “GPU killer” just months later is a classic micro-cap maneuver to chase retail liquidity.

Trailing twelve-month net losses sit at roughly $23.4 million with minimal revenue. The company filed a $100 million mixed securities shelf in February.

Shares traded around $4.46 Thursday, up 27% on the session but still down from a 52-week high of $44.40. Market cap is roughly $36 million.

When one user on X questioned the company’s small employee count and suggested a pump and dump, Shkreli pushed back directly: “I literally work for this company.”

Shkreli acknowledged in his essay that photonic systems still lack true optical memory and struggle with nonlinear operations between neural network layers.

Most engineers expect the nearer-term opportunity for photonics is moving data between GPUs, not replacing them for computation.

Image: IMAGN