a16z Data: Cheaper AI Tokens Drive Demand While H100 Rental Prices Hold

Cheaper AI is making the world hungrier, and H100 rental prices are not shrinking. Data from Ornn, Silicon Data and Bloomberg through August 2026 shows what a16z calls a textbook Jevons paradox in the AI market, according to The Decoder. Token prices keep dropping, but H100 GPU rental prices hold steady or climb.

The mechanism is simple. Cheaper tokens unlock agents, automation and new applications, so volume grows faster than per-unit costs fall. Who is doing all the consuming? Not clear. The report says how much demand comes from humans versus the systems themselves isn't clear, since agentic AI burns through tokens at a staggering rate. Compute demand could be artificially inflated, it adds, and even modest human usage growth could trigger outsized hardware needs.

Here is the context: the whole system rests on one assumption, that AI usage has to grow fast enough to offset falling token prices. If demand flattens, the chain from chip makers and memory suppliers to energy providers and cloud companies takes a hit. The Decoder points to US stocks dropping on reports that OpenAI's annualized revenue might be lower than previously reported. The piece, as relayed here, gives no figure or date for that drop, and no percentages for token prices or rental rates.

My take? Prices down, rents holding or climbing, an entire industry holding hands and betting on appetite. It is magnificent. But a boom that rests on a single assumption deserves a very, very good seat belt.