Global shipments of AI Server Compute ASICs (Application-Specific Integrated Circuits) for servers are projected to triple by 2027 compared to 2024 levels, according to the latest Data Center AI Server Compute ASICs Shipment Forecast and Tracker from Counterpoint Research’s HPC (High Performance Computing) Service.

This explosive growth is underpinned by robust demand for Google’s TPU infrastructure to support Gemini, sustained scaling of AWS Trainium clusters, and ramp-ups at Meta (MTIA) and Microsoft (Maia) as they expand their internal silicon portfolios.

Commenting on the growing data center AI Server Compute ASIC market, Research Vice President Neil Shah said, “In-house AI Server Compute ASIC design growth is validating the in-house custom XPU era, where AI accelerators are tailor-made for special and specific workloads (training or inference), structurally diversifying beyond solely relying on general purpose GPUs. As power and space become a bottleneck, moving some AI workloads to vertically integrated silicon gives hyperscalers more control and leverage, but it also comes with significant software plumbing to optimize the AI workloads and enjoy the power and performance benefits.”

Shah added, “The tides are changing fast, and we estimate that global data center AI Server Compute ASIC shipments will cross the 15-million mark in 2028, surpassing data center GPU shipments. The top 10 AI hyperscalers combined will deploy more than 40 million AI Server Compute ASIC chips cumulatively during 2024-2028. What is also supporting this unprecedented demand is AI hyperscalers building significant rack-scale AI infrastructure based on their in-house stacks, such as Google TPU Pods and AWS Trainium UltraClusters, enabling them to operate as one supercomputer.”

In terms of shipment volumes by AI hyperscaler, Google is projected to retain its lead in the market through 2027, driven by the explosive growth of its Gemini ecosystem.

Commenting on Google’s dominance, Research Associate David Wu highlighted, “Although Google’s market share is expected to fall to 52% in 2027 due to the expanding TAM and competing hyperscalers adopting internal silicon in partnership with design houses such as Broadcom, Marvell and Alchip, its TPU fleet will remain the undisputed volume backbone and north star of the industry. This baseline is underpinned by the massive and sustained compute intensity required for training and serving next-generation Gemini models, which necessitates a continuous, aggressive ramp-up of internal silicon infrastructure.”

The broader AI Server Compute ASIC market is undergoing a structural transformation, evolving from a concentrated duopoly, dominated by Google and AWS in 2024, into a diversified ecosystem. By 2027, the shipment mix will fragment to include volume ramp-ups from Meta (MTIA) and Microsoft (Maia). This shift highlights a strategic pivot across hyperscalers to decouple from merchant silicon (e.g. NVIDIA) and scale internal custom silicon to optimize performance-per-watt for their specific workloads. 

Commenting on the AI Server Compute ASIC design partner dynamics, Associate Director Brady Wang highlighted, “While Broadcom has been the number one choice as an AI Server Compute ASIC design partner, it is poised to face intensifying competitive pressure from the emerging Google-MediaTek alliance. This shift is most evident in the upcoming TPU v8 series, where Google is adopting a dual-sourcing strategy – while Broadcom maintains its lead on the high-performance TPU v8AX 'Sunfish' (training), MediaTek has secured a design partnership for the inference-focused TPU v8x ‘Zebrafish’.”

Although MediaTek’s high-volume ramp has faced an approximate one-year delay, its entry into the data center segment represents a formidable challenge to Broadcom's long-standing market dominance. Wang added, “While Broadcom continues to hold dominant IP moats and delivers industry-leading efficiency for complex training workloads, its premium pricing structure contrasts sharply with MediaTek’s more cost-effective approach. As Google aggressively scales its global infrastructure, it has become strategically imperative for the company to diversify its supply chain and leverage MediaTek’s lower-cost solutions for mass-deployment inference tasks.”

Taiwan’s Alchip is set to regain its market position as it expands into the AWS supply chain as one of the premier design partners for the coming years. With our supply chain checks indicating Alchip’s arrival, Marvell could face intensifying competitive headwinds related to the future AWS Trainium roadmap. With Microsoft’s Maia series being the major data center AI Server Compute ASIC ramp, Marvell would be aiming to diversify its pipeline. Marvell’s share is estimated to dip to 8% in 2027 as competition intensifies, but we still see the shipments doubling between 2024 and 2027.

Commenting on Marvell’s competitive positioning in the next few years, Associate Director Gareth Owen highlighted, “Securing new net-new design wins is now critical to de-risk its growth thesis and prevent a potential revenue air pocket after the current Trainium series. Having said that, Marvell’s end-to-end custom chip portfolio looks more solid than ever, with its custom silicon innovations, such as its customized HBM/SRAM memory and PIVR solutions, and the Celestial AI acquisition broadening Marvell’s addressable market in scale-up connectivity. Celestial AI could not only add multi-billion-dollar increments to Marvell’s revenues every year but also potentially drive a leadership position in optical scale-up connectivity in the coming years. We believe that these innovations will be critical for Marvell’s custom XPU and XPU attach opportunities going forward and improve its design share in the coming years.”

This research stems from our data center compute tracker and forecast triangulated with foundry (front-end and back-end) wafer consumption and capacity forecasts on the supply side and data center capex and AI applications growth on the demand side from our HPC service.

About Counterpoint Research

Counterpoint Research is a global market research firm specializing in products across the technology ecosystem. We advise a diverse range of clients – from smartphone OEMs to chipmakers and channel players to Big Tech – through our offices located in the world's major innovation hubs, manufacturing clusters and commercial centers. Our analyst team, led by seasoned experts, engages with stakeholders across the enterprise – from the C-suite to professionals in strategy, analyst relations (AR), market intelligence (MI), business intelligence (BI), product and marketing – to deliver services spanning market data, industry thought leadership and consulting. Our core areas of coverage include AI, Automotive, Consumer Electronics, Displays, eSIM, IoT, Location Platforms, Macroeconomics, Manufacturing, Networks and Infrastructure, Semiconductors, Smartphones and Wearables. Visit our Insights page to explore our publicly available market data, insights and thought leadership, and to understand our focus, meet our analysts and start a conversation.