Martin Shkreli Says New Photonic Chips Could Run Circles Around Nvidia

Martin Shkreli thinks the next serious challenge to NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) could run on light. On an All-In podcast episode released October 12, 2026, the former hedge fund manager named three private startups building optical chips and said their designs “would run circles around what Nvidia does.” The comment comes during a rough stretch for the stock. Nvidia shares fell 6.39% over the past week as chip stocks slid in a global tech rout on AI funding worries.

What Shkreli Said About Photonic Chips

Shkreli told the hosts he is exploring the field himself. “So we’re investigating, maybe we’ll start a company or something like that. But there are a couple of startups like Neurofos, Opticore, Lumi that want to make these photonic matmul cores which would run circles around what Nvidia does. Now of course that’s not the whole story. There are a lot of details and nonlinearity and storage.” He said the optical calculations “happen in instant time.”

Neurofos, Opticore and Lumi are private, so public investors have no direct way to own them and no disclosed financials to evaluate.

How Photonic Computing Works in Plain English

AI models mostly run on matrix multiplication (“matmul”): multiplying enormous grids of numbers over and over. A GPU does that math by pushing electrons through billions of transistors, which burns power and throws off heat. A photonic chip encodes numbers in beams of light and lets them interact as they pass through optical components, so the answer comes out as the light crosses the chip. That gives you the speed and energy appeal.

Shkreli’s cautions carry real weight. Neural networks also need nonlinear steps between multiplications (functions that decide which signals pass forward), and light handles those poorly. Models also need memory to hold their weights and data, and that storage remains electronic. Any photonic core still needs conventional silicon wrapped around it.

Nvidia Is Already Wiring Light Into Its Racks

Any competition faces a company at full speed. In Q2 FY2027, Nvidia reported revenue of $96.22 billion, up 105.8% year over year, with Data Center revenue of $89.02 billion, according to its earnings release filed with the SEC. Data Center Networking alone brought in $40.313 billion, up 138%.

Nvidia is also pulling optics into its own roadmap. It is shipping Spectrum-6 switch systems supporting pluggable and co-packaged optics, collaborating on silicon photonics with Marvell (NASDAQ:MRVL), and has signed multi-year optics agreements with Coherent (NYSE:COHR), Corning (NYSE:GLW) and Lumentum (NASDAQ:LITE). Today those optics move data between chips. The startups Shkreli named want light to do the math itself, which is a harder engineering problem. They would also have to compete with CUDA, the software platform that runs under Nvidia’s hardware. Chief executive Jensen Huang summed up the demand picture last quarter: “Now, compute is revenue.”

Shkreli’s Short on Quantum Computing

The hosts also asked about Shkreli’s “hatred for quantum” and what they called his “raging short that may or may not be winning.” The episode description frames his view as why quantum computing is a one-trick business.

On his background, Shkreli said he started at 16 at Jim Cramer’s hedge fund, said Cramer made 23% a year net of fees, and later moved to a Tiger Cub fund. When the hosts noted that he bought a drug and raised its price, he answered “Yes.” The episode also covered what prison taught him.

What Nvidia Investors Should Track Next

Nvidia carries a market cap of about $5.17 trillion and a P/E of 44. Shares closed at $214.63, up 15.35% year to date. Management guided Q3 revenue to $108.0 billion +/- 2% and disclosed supply commitments of $279 billion, mostly tied to memory for Vera Rubin.

Photonic matmul chips belong on the long-term radar. What matters near term is whether any of these startups shows benchmarks on full AI models, lands a hyperscaler customer, or solves the nonlinearity and memory problems Shkreli pointed out. Until that happens, the bigger questions for Nvidia holders are AI financing concerns and the Q3 report. The traits that showed up early in past tech monsters are the same ones we cataloged in a free playbook on spotting the next Nvidia. Investors can follow Nvidia’s own updates through its investor relations site.

Contact [email protected] for any questions or corrections.