While focus is on massive supercomputers for training next-gen models, the real supply chain constraint will be 'inference' chips—the GPUs needed to run models for billions of users. As adoption goes mainstream, demand for everyday AI use will far outstrip the supply of available hardware.

Templar's Sam Dare argues the perceived GPU scarcity is misunderstood. The actual bottleneck is the limited supply of the latest, well-connected GPUs in data centers. His project aims to create algorithms that can effectively utilize the vast, distributed network of consumer-grade and older enterprise GPUs, unlocking a massive new compute resource.

The widely discussed GPU supply crunch is only half the problem. There's a severe shortage of suppliers who can operate data centers with the high reliability and SLAs required for mission-critical inference. Out of many providers, only a handful meet the "gold tier" for operational excellence.

Anthropic is throttling user access during peak hours due to GPU shortages. This confirms that the AI industry remains severely compute-constrained and validates the multi-billion dollar infrastructure investments by giants like OpenAI and Meta, which once seemed excessive.

While NVIDIA's GPUs have been the primary AI constraint, the bottleneck is now moving to other essential subsystems. Memory, networking interconnects, and power management are emerging as the next critical choke points, signaling a new wave of investment opportunities in the hardware stack beyond core compute.

A critical, under-discussed constraint on Chinese AI progress is the compute bottleneck caused by inference. Their massive user base consumes available GPU capacity serving requests, leaving little compute for the R&D and training needed to innovate and improve their models.

While many focus on compute metrics like FLOPS, the primary bottleneck for large AI models is memory bandwidth—the speed of loading weights into the GPU. This single metric is a better indicator of real-world performance from one GPU generation to the next than raw compute power.

GPUs were designed for graphics, not AI. It was a "twist of fate" that their massively parallel architecture suited AI workloads. Chips designed from scratch for AI would be much more efficient, opening the door for new startups to build better, more specialized hardware and challenge incumbents.

Previously, the biggest constraint in AI was compute for training next-gen models. Now, the critical bottleneck is providing enough compute for *inference*—the real-time processing of queries from a rapidly growing user base.