Gimlet Labs just raised $80 million to solve one of AI’s biggest headaches. Their technology lets artificial intelligence run on any computer chip, whether it’s made by NVIDIA, AMD, Intel, or newer players like Cerebras.
Right now, most AI systems are picky about hardware. If you build your AI for NVIDIA chips, it won’t easily work on AMD chips. This creates expensive bottlenecks and limits how fast AI can grow.
Breaking Down AI’s Chip Prison
Think of it like having a car that only works on certain roads. Gimlet’s technology is like building a universal car that can drive on any surface. Companies can mix and match different chips based on what’s available and affordable, rather than being locked into one expensive option.
The timing couldn’t be better. NVIDIA’s chips are backordered for months, and companies are desperate for alternatives. Meanwhile, new chip makers like Cerebras and d-Matrix are creating powerful alternatives, but they’ve been hard to use alongside existing systems.
Gimlet’s approach lets AI models automatically split their work across whatever chips are available. One part might run on an NVIDIA chip while another part runs on AMD hardware, all working together seamlessly.
What This Means
This could dramatically reduce AI costs and speed up development. Instead of waiting months for specific chips or paying premium prices, companies could use whatever computing power they can get their hands on. For businesses trying to add AI features, this means faster deployment and lower costs.