Advanced Micro Devices IncSecured large-scale inference deals with OpenAI, Meta, Microsoft, and Anthropic.
Bloomberg Intelligence projects the AI inference market will grow at a 32% annual compound rate through 2032, reaching $1.3 trillion, roughly double the $658 billion it forecasts for the AI training market over the same period. Nvidia has brought Groq and its SRAM-based language processing units into its fold, pairing them with its HBM-packaged GPUs to handle both the decode and pre-fill phases of inference. Advanced Micro Devices uses a chiplet design that allows its GPUs to be packaged with more memory, and it has secured large-scale inference deals with OpenAI, Meta Platforms, Microsoft, and Anthropic, while also partnering with Cerebras to incorporate its SRAM-based solution into AMD's Helios racks. Cerebras produces massive wafer-sized chips with embedded SRAM and is breaking into the mainstream with deals with OpenAI, while its partnership with AMD will help it adopt a lower-cost system. Broadcom helps hyperscalers turn custom AI chip designs into physical chips packaged with HBM, seeing this as a more than $100 billion business in fiscal 2027, and SK Hynix, the leader in HBM with a nearly 60% market share, recently signed a $500 billion deal with Nvidia.
Advanced Micro Devices IncSecured large-scale inference deals with OpenAI, Meta, Microsoft, and Anthropic.
NVIDIA CorporationInference market projected to reach $1.3 trillion by 2032; Nvidia integrates Groq and signs $500 billion HBM deal with SK Hynix.
Broadcom IncHyperscaler custom AI chip business seen as over $100 billion in fiscal 2027.
Cerebras Systems Inc. Class A Common StockBreaking into mainstream with OpenAI deals and AMD partnership.
Meta Platforms Inc.
Microsoft Corporation
SK Hynix IncLeader in HBM with nearly 60% market share; signed $500 billion deal with Nvidia.
OpenAI is a major customer for inference deals with AMD and Cerebras, benefiting from market growth.