Major build-outs run in two phases, per Carlota Perez: the infrastructure, and then the applications that pay for it. The internet took seven years to close the gap, and AI is closing it in about half the time. Whether that is circular depends on how large the circle is.
Yesterday brought significant data to assess precisely the health of inference business models, & the data suggests they are quite healthy, even before the pricing ladder has room to climb.
Tokenized real-world assets are compounding at 53% a month, more than twice as fast as Anthropic. Stocks & commodities now trade on blockchains 24/7, powered by Allium, which now runs Bloomberg's RWA terminal dashboard.
GPU rental prices doubled in six months while inference prices kept falling. The reconciliation is efficiency : the hinge that turns scarce silicon into cheap intelligence.
NVIDIA guided Q3 to $108b, but hyperscaler growth slowed to 13% sequentially. The buyers making up the difference need financing, so DSO jumped 15 days to 60 & the commitment stack reached $581b including a $101b equity portfolio in its own customers.
AI infrastructure shortages do not hit simultaneously. They cascade in multi-year waves across the server rack & into the physical grid : GPUs in 2023, memory & flash in 2024–2025, CPUs in late 2025, spinning disk in 2026, & data centers at $20b per gigawatt.
SpaceXAI spent $18b of capital last quarter, $16b of it on AI, a hyperscaler's budget covered 12% by operations. Equity & credit have both marked it down.
Ollama has become the largest developer platform for open models, with 8.9m developers & one million new installs each week. Today we announced our investment in the Series B.
Prioritize routing over model choice. Most AI work runs on cheap local models.
The inference market is the largest in software. We invested in Sail's Series A to build queuing infrastructure for long-horizon agents that run for hours.
