Musk Warns AI Industry May Face Hundreds of Billions in Funding Gap Next Year
Keras creator François Chollet argues GPU spot prices keep rising, HBM and DRAM capacity is booked out to 2029, frontier training compute doubles roughly every five months, and inference margins can reach 70-85%. Musk replied that powering, cooling and stabilizing such compute is extremely hard, warning that a few months' delay in revenue generation could require hundreds of billions in extra working capital industry-wide, and asked where that money would come from.
Related Entities
Full Report
IT之家 November 22 news: François Chollet, founder of the open-source deep learning project Keras, posted on X saying that either you believe AI infrastructure buildout is over, that AI is just a fleeting bubble, and that compute demand has peaked; or, this is a rare buying opportunity.
François stated that GPU spot prices are still rising, production capacity is hard to guarantee, HBM and DRAM capacity through 2029 is basically fully booked, and chip and memory supply is expected to remain tight, continuing into the 2030s. Training compute for frontier models doubles roughly every 5 months, and model capability improvement is more than twice as fast as before the emergence of reasoning models. AI inference business margins can reach 70% to 85%, demand is still growing rapidly, and each new generation of models brings new application scenarios.

IT之家 noted that Musk replied to François, saying that powering, cooling, and achieving stable operation of such large-scale compute equipment is extremely difficult.
(The AI industry) may face a severe funding shortage next year. If across the entire industry the time from paying compute costs to equipment being able to generate stable revenue is delayed by several months in aggregate, working capital alone may require hundreds of billions of dollars in additional investment. Where will that money come from?
