Future Teslas will contain powerful AI inference chips that sit idle most of the day, creating an opportunity for a distributed compute network. Owners could opt-in to let Tesla use this power for external tasks, earning revenue that offsets electricity costs or the car itself.

George Hotz outlines a contrarian AI infrastructure strategy. Instead of expensive enterprise hardware, Tiny Corp plans to use upcoming consumer AMD GPUs, pair them with extremely cheap power in Oregon (~$0.03/kWh), and sell compute tokens on existing platforms. This low-overhead model aims to undercut traditional cloud providers.

The vast network of consumer devices represents a massive, underutilized compute resource. Companies like Apple and Tesla can leverage these devices for AI workloads when they're idle, creating a virtual cloud where users have already paid for the hardware (CapEx).

Projects like BitTensor represent a fundamental threat to the centralized, capital-intensive AI labs. By distributing the model training process via open-source orchestration, they offer an "orthogonal attack vector" that could democratize AI if capital markets stop writing multi-billion dollar checks for compute.

While AI inference can be decentralized, training the most powerful models demands extreme centralization of compute. The necessity for high-bandwidth, low-latency communication between GPUs means the best models are trained by concentrating hardware in the smallest possible physical space, a direct contradiction to decentralized ideals.

IOTA's technology is designed to work with compute that can be taken away at a moment's notice. This allows it to acquire unused data center time for as little as 10 cents on the dollar—a resource no traditional, synchronous training method can utilize.

The current focus on building massive, centralized AI training clusters represents the 'mainframe' era of AI. The next three years will see a shift toward a distributed model, similar to computing's move from mainframes to PCs. This involves pushing smaller, efficient inference models out to a wide array of devices.

Block's CTO believes the key to building complex applications with AI isn't a single, powerful model. Instead, he predicts a future of "swarm intelligence"—where hundreds of smaller, cheaper, open-source agents work collaboratively, with their collective capability surpassing any individual large model.

Templar's decentralized AI training model doesn't require specific GPUs. Instead, it defines the validation criteria for a correct output. This forces miners to find the most economically efficient hardware and software combination to solve the problem, a process Sam Dare calls "emergence," where optimal solutions arise from the incentive structure itself.