• NVIDIA has publicly refuted claims of ongoing severe shortages for its flagship H100 and H200 AI GPUs.
  • The company states that supply chain improvements have slashed delivery times from nearly a year to just 8-12 weeks.
  • Despite improved availability, underlying demand from AI firms and cloud providers remains historically intense, keeping unit costs elevated.

NVIDIA has moved to quell mounting industry speculation, issuing a firm denial that its critical H100 and H200 data center GPUs are stuck in a prolonged shortage. The semiconductor giant asserts that significant improvements within its supply chain have dramatically increased availability and reduced wait times for customers.

According to people familiar with the matter, real-world delivery times for the high-demand H100 units, which had stretched to an arduous 8–11 months throughout much of 2023 and 2024, have now normalized to a range of 8–12 weeks. This easing of constraints is a direct result of increased manufacturing and advanced packaging capacity, primarily at key foundry partner TSMC, which has been rapidly scaling its CoWoS packaging technology essential for these chips.

The denial comes amid record-shattering financial performance for NVIDIA. The company's most recent quarterly report showed data center revenue, which is overwhelmingly driven by these AI GPUs, surging to $30.7 billion. This unprecedented demand, fueled by a global race to build and deploy AI infrastructure, had indeed created a historic supply crunch. The cost for a single H100 GPU remains high, often ranging between $25,000 to $35,000, a reflection of both the complex supply chain and the unrelenting demand.

An NVIDIA spokesperson, when reached for comment, reiterated that the company has seen "marked improvements in our ability to meet customer demand" and that the situation is continuously evolving. Major cloud providers and server OEMs, who had been allocating scarce GPU resources carefully, are now reportedly seeing more consistent shipment schedules, allowing for more predictable deployment of AI services.

However, analysts caution that the market remains exceptionally tight by historical standards. While the most acute shortage phase may be over, the underlying demand dynamics have not fundamentally changed. The launch of NVIDIA's next-generation Blackwell architecture later this year could test the resilience of these supply chain improvements once again.

In a related industry shift, the intense focus on producing data center AI chips has had a knock-on effect, worsening the supply situation for NVIDIA's consumer-grade GeForce gaming GPUs. This has led to frustration among gamers and higher prices in that segment, highlighting the company's strategic prioritization of its immensely profitable AI business.

This article was updated to clarify the current delivery timeframes for H100 GPUs.