Google unveiled two AI ASIC server compute chips at the recently concluded Google Cloud Next – TPU v8t (training) and TPU v8i (inferencing). These are the eighth generation of Google’s TPU chips designed for scalability, efficiency and reliability to support the growing reasoning-heavy workloads and MoE architectures for the Agentic AI era. As highlighted in Counterpoint Research’s latest edition of Foundry Quarterly and Monthly Intelligence report, the TPU v8t (Zebrafish) holds a pivotal position in Google's AI silicon strategy. We view this generation as a critical inflection point from a supply chain perspective, as it marks the first major step in Google's diversification away from a pure Broadcom turnkey ASIC model.

On the primary motivation behind Google's adoption of the MediaTek model, where Google designs the compute die and MediaTek provides the I/O die, Research Associate David Wu said, “It lies in HBM procurement economics. Under the Broadcom turnkey model, the ASIC vendor handles HBM sourcing and applies a 15%-20% markup. As HBM represents an increasingly large share of the total ASIC bill of materials, this markup becomes a high cost at scale, particularly as Google's TPU deployment velocity continues to accelerate. By bringing chip design and HBM procurement in-house, starting with Zebrafish, Google eliminates the intermediary fee and lowers its overall cost.”

MediaTek is set to experience substantial volume growth as Zebrafish commences production in late 2026 and Humufish scales through 2028. Based on our latest Global AI Server Compute ASIC Shipment Forecast, we project combined TPU v8t and v8e shipments to approach 5 million units in 2028, more than 10x growth from the approximately 400k units shipped in 2026, driven by Google’s accelerating TPU deployment across both internal workloads and Cloud customers.

Commenting on the outlook, Associate Director Brady Wang said, “This trajectory positions MediaTek to capture around 26% of total AI server compute ASIC shipments by 2028. It is important to note that our current projections do not include potential project wins for Meta MTIA. Besides, achieving the projected volume is contingent upon sufficient advanced packaging capacity, including TSMC CoWoS allocation and Intel EMIB-T yield readiness for Humufish.”

A major execution risk lies in the TPU v8e (Humufish) roadmap, where MediaTek is proposing Intel’s EMIB-T packaging. Senior Analyst Ashwath Rao said, “Currently in the design-in and qualification phase, with mass production not targeted until late 2027, this transition introduces highly specific execution risks. Key variables include the yield ramp of Intel Foundry Services (IFS) advanced packaging and the readiness of critical substrate vendors, which could ultimately impact MediaTek's volume delivery trajectory.”

In our latest Foundry Quarterly and Monthly Intelligence report, we also address:

1. Recent engineering updates on the TPU SerDes I/O die modifications and running changes.
2. Marvell's expanding custom silicon engagement with Google.
3. An evaluation of whether the TPU v8e (Humufish) will exclusively rely on Intel EMIB-T packaging.

For detailed research, refer to the following reports available to subscribers:

• Arm-based CPUs to Capture 90% of AI ASIC Server CPU Market by 2029
• Global AI Server Compute ASIC CPU Attach Shipment by Chip Model, Hyperscalar, Design Partner & Foundry Forecast, Mar 2026
• HBM Demand for AI Server Compute ASICs to Grow 35x by 2028

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.