Mythos, Muse, and the Opportunity Cost of Compute

The Stratechery argument quietly inverts the standard AI moat thesis. Most coverage focuses on who controls model supply; this piece asks whether demand aggregation is the more durable position when compute scarcity makes supply-side control unreliable.

This connects directly to two threads running through recent Modelwire coverage. The MIT Technology Review piece on 'treating enterprise AI as an operating layer' (April 16) makes a structurally similar argument from the enterprise side: the competitive position lives in the deployment and governance infrastructure, not the model itself. That's a demand-side control argument dressed in operational language. Meanwhile, The Verge's reporting on the RAM shortage (April 18) provides the material grounding Stratechery's thesis needs: if DRAM suppliers are projected to meet only 60% of global demand by end-2027, compute scarcity isn't a theoretical constraint, it's an active one. Together, these stories suggest a coherent picture where both infrastructure bottlenecks and demand aggregation are being repriced simultaneously, which makes the timing of Stratechery's Aggregation Theory revisit more than coincidental.

Watch whether OpenAI's consumer app acquisition strategy (flagged in the 'tokenmaxxing' piece from April 17) produces measurable demand concentration metrics within the next two quarters. If OpenAI can demonstrate that app ownership translates to preferential compute allocation during shortage periods, Stratechery's demand-control thesis gets its first real-world test case.