The AI market is becoming "polytheistic," with numerous specialized models excelling at niche tasks, rather than "monotheistic," where a single super-model dominates. This fragmentation creates opportunities for differentiated startups to thrive by building effective models for specific use cases, as no single model has mastered everything.

The AI ecosystem will evolve into an "orchestration age" where large 'boss' models delegate tasks to a network of smaller, faster, specialized models. This means different chip architectures (e.g., NVIDIA for large models, Cerebras for speed) will function as complementary parts of a larger system, not just direct competitors.

Just as developers use various databases for different needs, AI applications will rely on a "constellation" of specialized models. Some tasks will require expensive, high-reasoning models, while others will prioritize low-latency or low-cost models. The market will become heterogeneous, not monolithic.

The current AI boom focuses on GPUs for "thinking" (Gen AI). The next phase, "Agentic AI" for "doing," will rely heavily on CPUs for task orchestration and memory for context, creating new investment opportunities in this previously overshadowed hardware.

Enterprises will shift from relying on a single large language model to using orchestration platforms. These platforms will allow them to 'hot swap' various models—including smaller, specialized ones—for different tasks within a single system, optimizing for performance, cost, and use case without being locked into one provider.

OpenAI's compute deal with Cerebras, alongside deals with AMD and Nvidia, shows that hyperscalers are aggressively diversifying their AI chip supply. This creates a massive opportunity for smaller, specialized silicon teams, heralding a new competitive era reminiscent of the PC wars.

The rise of agent orchestration using specialized, open-source models will drive demand for custom ASICs. Jerry Murdock argues that putting a model on a dedicated chip will be far cheaper and more tunable for specific workloads than using expensive, general-purpose GPUs like Nvidia's, spurring a hardware shift.

The inference market is too large to remain monolithic. It will fragment into specialized platforms for different use cases like real-time video, long-running agents, or language models. This specialization will extend to hardware, with high-throughput, low-latency-need tasks (like agents) favoring cheaper AMD/Intel chips over NVIDIA's top GPUs.

Beyond the simple training-inference binary, Arm's CEO sees a third category: smaller, specialized models for reinforcement learning. These chips will handle both training and inference, acting like 'student teachers' taught by giant foundational models.