Martin Shkreli claims photonic chips could outperform Nvidia

By Splitfeed EditorialLast updated Oct 19, 2026, 02:30 UTC

On October 17, 2026, Martin Shkreli made public claims that photonic chips could outperform Nvidia in AI processing tasks, with reports appearing simultaneously in 24/7 Wall St. and Yahoo Finance. Two days later, on October 19, a report from 36Kr argued that consumer home TV boxes are already capable of running large AI models, framing the question as one of practical alternatives to Nvidia hardware. Neither set of reports named specific chip manufacturers, provided benchmark figures, or cited technical specifications to support either claim.

The context that gives these stories weight is Nvidia's near-monopoly grip on AI accelerator hardware, which has made it a central bottleneck for AI development costs and accessibility. Photonic computing, which uses light rather than electrical signals to process data, has long been discussed as a potential leap forward in speed and energy efficiency, but it remains largely pre-commercial. The TV box claim points in a different direction, suggesting that software optimization and quantization techniques may allow commodity hardware to do work once reserved for expensive GPU clusters. What is genuinely in dispute is whether either path represents a near-term competitive threat to Nvidia or remains, for now, in the category of promising but unproven alternatives.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, have published reactions to these specific reports. Pro-AI voices would typically welcome both claims as evidence that AI capability is becoming cheaper and more distributed, democratizing access beyond well-funded labs. Anti-AI commentators would likely treat the spread of AI to low-cost consumer hardware with concern, arguing it accelerates deployment faster than safety or regulatory frameworks can keep pace. A Middle Ground perspective would probably call for independent benchmarking before reading too much into claims associated with a figure as controversial as Shkreli, while acknowledging that hardware diversification is a legitimate and important trend worth monitoring.

The clearest thing to watch is whether any photonic chip company surfaces with verifiable benchmarks against current Nvidia hardware, and whether the TV box claims hold up under independent testing with named models and measurable performance metrics. Those results would move the conversation from assertion to evidence.