Science and Technology · Kalshi ·June 24, 2026 ·Edited by Samuel Jo

Nvidia H200 compute pricing faces Chinese pause amid rising AI costs

Nvidia is confronting a potential price surge for its H200 AI accelerator, with the compute‑per‑hour rate projected to exceed $1.99 by June 30, 2026, as Chinese manufacturers pause new orders. The H200, positioned as the next‑generation successor to the popular H100, has become a focal point for AI developers seeking higher throughput. Industry observers note that China’s directive to hold off on H200 purchases has already strained supply chains, while Broadcom’s recent unveiling of a custom OpenAI chip intensifies competitive pressure. Traders currently assign a 99% probability to the price breaching that level, reflecting confidence in the upward pressure.

Background

The H200 chip, launched by Nvidia in early 2025, promises up to 30% more tensor performance than its H100 predecessor, making it attractive for large‑scale language model training. However, a June 2025 notice from Chinese regulators urged domestic AI firms to suspend further H200 acquisitions, citing concerns over export controls and domestic chip development. This pause coincides with a reported 20‑30% jump in H100 rental rates across Chinese data centers, indicating strong demand for Nvidia’s hardware despite the restriction. At the same time, Broadcom announced a custom silicon solution for OpenAI, signaling a potential shift in the AI compute landscape and putting additional pricing pressure on Nvidia’s offerings.

What to watch

Watch for the market’s resolution on June 30, 2026, when the H200 compute price will be confirmed. Any shift in Chinese policy toward the H200 or a formal launch of Broadcom’s OpenAI chip before that date could materially affect pricing dynamics. A significant move in the 0 pts odds would also signal changing market sentiment.

AI-written briefing grounded in 5 sources and the live market, edited by Samuel Jo. Odds are crowd probabilities, not advice — how this works.