Tom's Hardware H100 projections

web

Tom's Hardware·tomshardware.com/news/nvidia-to-reportedly-triple-output-...

Credibility Rating

3/5

Good(3)

Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.

Rating inherited from publication venue: Tom's Hardware

Relevant for tracking AI compute supply trends; H100 GPU availability is a key constraint on frontier model training and a focal point in AI governance discussions around compute thresholds.

Metadata

Importance: 38/100news articlenews

Summary

Nvidia planned to significantly scale H100 GPU production in 2024, potentially tripling output to up to 2 million units to meet surging AI and HPC demand. This expansion reflects the critical role of high-end compute hardware in enabling frontier AI development, though production scaling faces technical and supply chain challenges.

Key Points

•Nvidia reportedly aimed to produce up to 2 million H100 compute GPUs in 2024, roughly tripling prior output levels.
•Massive demand from AI training and HPC applications was the primary driver behind the aggressive production scale-up.
•Technical and manufacturing challenges posed risks to meeting the projected production targets.
•H100 availability is a key bottleneck for frontier AI development, making production projections strategically significant.
•This ramp-up signals growing industrial investment in AI compute infrastructure at unprecedented scale.

Review

The source details Nvidia's ambitious plan to dramatically scale up production of its H100 compute GPUs, a critical component for AI and high-performance computing. The company aims to increase output from approximately 500,000 units in 2023 to between 1.5 and 2 million units in 2024, representing a threefold increase that could generate substantial revenue. The production scaling faces several technical challenges, including the complex manufacturing of the large 814 mm² GH100 processor, securing sufficient 4N wafer supply from TSMC, obtaining HBM memory packages, and ensuring partner capacity for AI server production. Despite these obstacles, the massive demand for Nvidia's CUDA-based GPUs from major cloud providers like Amazon and Google underscores the strategic importance of this expansion. The potential success of this plan could significantly reshape the AI computing landscape and cement Nvidia's leadership in AI infrastructure.

Cached Content Preview

HTTP 200Fetched Apr 9, 202615 KB

Nvidia to Reportedly Triple Output of Compute GPUs in 2024: Up to 2 Million H100s | Tom's Hardware 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 
 

 

 

 

 

 

 
 
 
 

 
 

 Skip to main content 
 

 
 
 
 Tom's Hardware Subscription Why subscribe? Get deeper insights with deeper News Analysis posts
 Read exclusive subscriber-only features and interviews
 Unlock access to Bench, our custom benchmark test visualizer, and compare products
 From $7 /mth Subscribe now 
 
 
 
 
 
 
 
 
 
 
 
 
 Don't miss these 
 
 
 
 
 
 
 

 

 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 DRAM 
 
 Micron's $24 billion Singapore fab could need 500 transformers, more than double the output of any single manufacturer 
 
 
 

 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Artificial Intelligence 
 
 Price of Nvidia's Vera Rubin NVL72 racks skyrockets to as much as $8.8 million apiece, but server makers' margins will be tight 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 GPUs 
 
 Huawei unveils new "Atlas 350" AI accelerator with 1.56 PFLOPS of FP4 compute & up to 112 GB of HBM 
 
 
 

 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Semiconductors 
 
 How Nvidia's $20 billion Groq 3 LPU deal reshapes the Nvidia Vera Rubin Platform 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 GPUs 
 
 Nvidia demonstrates Rubin Ultra tray, the world's first AI GPU with 1TB of HBM4E memory 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Artificial Intelligence 
 
 Examining Nvidia's 60 exaflop Vera Rubin POD &mdash; how seven chips underpin company's 40 rack AI factory supercomputer 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 GPUs 
 
 Nvidia updates data center roadmap with Rosa CPU and stacked Feynman GPUs 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Tech Industry 
 
 Nvidia market share in China falls to less than 60% 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 SSDs 
 
 Kioxia announces new Super High IOPS SSD that helps accelerate AI workloads on Nvidia GPUs 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Tech Industry 
 
 Jensen says Nvidia has received orders from Chinese customers for H200 GPUs, licenses from US gov't &mdash; H200 manufacturing restarting 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 GPUs 
 
 Nvidia launches Vera Rubin NVL72 AI supercomputer at CES 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Tech Industry 
 
 A deeper look at the tightened chipmaking supply chain, and where it may be headed in 2026 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 CPUs 
 
 Nvidia will only produce one 88-core Vera CPU model &mdash; Jensen says the company will make billions of dollars from a single SKU 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 GPUs 
 
 Nvidia announces Vera Rubin Space Module 
 
 
 

 
 
 
 
 
 
 
 

 
 
 
 
 
 
 

 
 
 
 Mini PCs 
 
 Nvidia DGX Spark gets 18% price increase as memory shortages bite 
 
 
 

 
 
 
 
 
 

 
 

 
 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 

... (truncated, 15 KB total)

Resource ID: 8bc7e77e73324df4 | Stable ID: sid_f9O0LW4WTX