Marvell Rallies 5% as Google Plans to Diversify for AI Chips

2026-04-21 · Source: AI Magazine · Field: Business & Management — Corporate Strategy & Leadership, Operations & Process Management, Emerging Technologies & Innovation · Depth: Novice, quick

Summary

Marvell Technology is reportedly in discussions with Alphabet's Google to develop new custom AI chips, a move that caused Marvell's shares (MRVL.O) to climb over 5% on Monday, April 20, reaching a 52-week peak. This potential partnership aims to diversify Google's custom chip suppliers beyond Broadcom, with talks possibly involving two distinct chips: a memory processing unit to complement Google's Tensor Processing Unit (TPU) and a new TPU specifically for running AI models. The broader trend sees tech giants investing in custom AI chips to reduce dependence on external suppliers like NVIDIA, seeking more cost-effective and energy-efficient solutions for daily AI operations. Marvell, which also received a US$2bn investment from NVIDIA in March 2026, anticipates its revenue to approach US$15bn by 2028.

Key takeaway

For CTOs and Directors of AI/ML evaluating chip procurement strategies, Google's reported talks with Marvell signal a critical shift towards multi-sourcing custom AI silicon. You should assess your organization's reliance on single-vendor solutions and explore partnerships for custom chip development to enhance cost efficiency, energy performance, and supply chain resilience for your AI infrastructure.

Key insights

Diversifying custom AI chip suppliers reduces costs and supply chain risks for major tech companies.

Principles

• Custom chips offer cost and energy efficiency for repetitive AI tasks.
• Multi-sourcing partnerships secure AI supply chains.
• Diversification mitigates technological and supply chain risks.

Method

Companies like Google are developing custom AI chips, such as TPUs, through partnerships with designers like Broadcom and potentially Marvell, to optimize AI model training and inferencing.

In practice

• Explore custom silicon for specific AI workloads.
• Evaluate multiple chip design partners.
• Integrate bespoke silicon into existing data centers.

Topics

• Marvell Technology
• Google AI Chips
• Custom Silicon
• Tensor Processing Unit
• AI Infrastructure

Best for: CTO, Director of AI/ML, Entrepreneur, Investor, Executive, VP of Engineering/Data

Related on AIssential

• Google plans nearly two million new AI chips as it turns to Marvell for custom designs — Google is developing custom AI chips with Marvell to enhance inference capabilities and diversify its supply chain.
• Google and Nvidia Consider Intel as Backup Chip Manufacturer — Google and Nvidia are exploring Intel for chip manufacturing diversification.
• Meta to Spend Billions on AMD Gear, AI Scare Trade Continues — Diversification in AI chip procurement and specialized AI agents are reshaping market dynamics and software valuations.
• ALERT: Google Eyes Samsung for Icefish TPU as AI Chip Demand Surges — Google is developing Icefish AI chips and exploring Samsung as a manufacturer to diversify its supply chain amid surging demand.
• 😺 Google ran out of cloud — Hyperscalers are investing heavily in AI infrastructure, with demand outstripping current supply and shifting competitive dynamics.
• 615: Nvidia Q4, Ben Evans on AI Economics, Meta + Google TPUs, China Solar at 1TW/Year, Anthropic vs DoD, AWS Outages, METR, Sora, and Marty Supreme — Diversifying AI compute suppliers and architectures is crucial for large-scale AI development, despite integration challenges.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI Magazine.