BlockBeats news, June 22: JPMorgan stated that as large cloud computing companies and tech giants seek to reduce AI computing costs, improve energy efficiency, and move away from reliance on a single path of general-purpose GPUs, the custom chip ASIC market is entering a new growth cycle, with Broadcom and Marvell expected to be the primary beneficiaries of this trend.
In a recent semiconductor industry report, JPMorgan analysts Harlan Sur and Mayur Ramdhani estimated that the digital AI ASIC market will reach approximately $60 billion to $70 billion by 2026, maintaining a compound annual growth rate of over 40% to 50% in the coming years. The report states that Broadcom currently holds about 80% to 85% of the high-end ASIC market, with Marvell ranking second at approximately 10% to 12%.
The rapidly growing demand for AI computing is transforming chip procurement structures. J.P. Morgan believes that customers such as Google, Amazon, Meta, Microsoft, OpenAI, and SoftBank/Arm are accelerating their development of proprietary or customized AI processors to achieve better performance, power efficiency, and total cost of ownership. Unlike NVIDIA and AMD’s general-purpose GPUs, ASICs are typically designed for specific customers, software stacks, or platforms, making them better suited for hyperscale cloud providers with large-scale internal workloads.
The report anticipates Broadcom's AI revenue to surge from approximately $20 billion in fiscal year 2025 to over $60 billion in fiscal year 2026, and reach more than $150 billion by fiscal year 2027. Its pipeline includes Google TPU, Meta MTIA, ByteDance AI video and networking chips, OpenAI XPU, SoftBank/Arm XPU, and Anthropic-related TPU rack-scale solutions.
For Marvell, J.P. Morgan expects its data center revenue to increase from approximately $6.1 billion in 2025 to approximately $9.3 billion in 2026, and reach approximately $14.6 billion in 2027. Growth drivers include Amazon Trainium 3 and Trainium 4, Microsoft Maia, Google SmartNIC/DPU, CXL controllers, and 800G/1.6T optical DSPs, coherent lite, and early CPO solutions.
The report also presents a key insight: by 2027, annual shipments of AI ASICs/XPUs will surpass those of GPUs. JPMorgan expects total AI accelerator shipments in 2027 to reach 23.3 million units, with 10.9 million GPUs (47%) and 12.5 million ASICs/XPUs (53%). This indicates that although GPUs will continue to grow, custom chips may capture a larger share of new AI compute deployments.
JPMorgan Chase cited Google/Broadcom’s TPU7x Ironwood and Nvidia’s Blackwell as examples showing that AI ASICs are competitive in terms of cost-performance and power efficiency. The report indicates that the TPU7x Ironwood offers FP8 performance close to that of Nvidia’s B200/B300, with an estimated price of approximately $13,000—lower than the B200’s $35,000 and the B300’s $40,000—and delivers superior compute per dollar and compute per watt compared to the reference GPUs.
This assessment does not imply that demand for Nvidia will quickly decline. Instead, it points to a divergence in AI infrastructure investment: GPUs will continue to serve general-purpose training and inference needs, while cloud providers' custom ASICs will achieve higher penetration in large-scale, stable, and predictable internal workloads.
For investors, J.P. Morgan’s report reinforces the logic that the AI hardware supply chain is expanding from GPUs to ASICs, advanced packaging, HBM interfaces, SerDes, optical interconnects, and CPO. If the report’s predictions materialize, Broadcom and Marvell will no longer be merely suppliers of AI networking or connectivity chips, but will become core platform companies in the next phase of AI computing architecture transition.