The AI market is shifting from a focus on compute scale to context, as institutional questions are structural rather than linguistic. Amazon, Alphabet, Meta and Microsoft together guide to roughly $725 billion of AI capital expenditure in 2026, up about 77 per cent on the prior year. Alphabet alone spent $44.9 billion in a single quarter, double the year before. Microsoft has guided its next fiscal year to $255 billion to $260 billion. In the last week of July, more than a trillion dollars came off the chip complex in five sessions, with Nvidia losing $238 billion of market value, SK Hynix $176 billion, Samsung $173 billion and Micron $113 billion, and AMD and TSMC each shedding more than $100 billion. The stated cause was a repricing of expectations on concern that infrastructure spending is peaking faster than revenue. Scaling a language model improves fluency and reasoning on problems resembling training data but does not create new information. Questions like routing positions through a lithography vendor or identifying shared funding sources require traversing a sourced, time-aware structure, not language fluency. A graph where every edge carries a source and a date provides answers that can be walked backwards, unlike a frontier model’s plausible but unverifiable output. The argument is that past a certain point, the return on another order of magnitude of compute is smaller than writing the domain down properly.