The industry is fixated on the GPU shortage, but the proliferation of AI agents will create massive demand for general-purpose compute, leading to a CPU bottleneck. As millions of agents perform tasks, the availability of CPU cores—not just specialized processors—will become the primary constraint on growth for compute providers.

The performance gains from Nvidia's Hopper to Blackwell GPUs come from increased size and power, not efficiency. This signals a potential scaling limit, creating an opportunity for radically new hardware primitives and neural network architectures beyond today's matrix-multiplication-centric models.

Jensen Huang emphasizes that Moore's Law is dead as a primary performance driver. The 50x gain from Hopper to Blackwell came overwhelmingly from architecture and computer science breakthroughs, with raw transistor improvements providing only marginal benefit.

Huawei is shifting from shrinking transistors (Moore's Law) to optimizing data flow via advanced chip stacking and interconnects. This "tau scaling law" is an innovative workaround to physical limits, aiming to create competitive AI compute power without access to the most advanced manufacturing processes.

As performance gains from general-purpose CPUs stalled, the industry shifted to domain-specific architectures (DSAs). By designing hardware like GPUs and TPUs for narrow tasks like AI, architects can achieve dramatic performance improvements that are no longer possible with traditional CPUs.

For two decades, silicon chips have been thermally constrained to a power density of about 1 watt per square millimeter. New R&D efforts are finally overcoming this barrier, which could lead to smaller, more powerful chips, despite significant thermal and electrical engineering challenges.

Adding more FLOPS to current AI chips is useless due to thermal throttling. Etched realized the solution is lowering voltage, which quadratically reduces power consumption. Inspired by bitcoin miners, they created a new power delivery system enabling chips to run at under half the voltage of GPUs.

While power supply is a current data center bottleneck, a more significant long-term risk is technological disruption. Chip innovations promising 10-1000x more power efficiency could make today's massive, power-centric data center investments obsolete or oversized before they are fully utilized.

Cerebras CEO Andrew Feldman claims that new AI chip architectures are breaking from the traditional 18-month doubling cycle of Moore's Law. Unlike mature GPU designs that rely on smaller manufacturing nodes for gains, new architectures have significant room for optimization, promising performance improvements far greater than 2x in the next cycle.