Google to Release New AI Chips, Challenging Nvidia

2026-04-20 · Source: Bloomberg Tech · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Capital Markets & Investment Management, Emerging Technologies & Innovation · Depth: Intermediate, extended

Summary

Bloomberg Tech discusses Google's strategic shift in its Tensor Processing Unit (TPU) program, moving towards specialized AI chips for inference rather than combined training and inference. This development, including potential collaborations with Marvell, aims to address the rapidly growing demand for AI inference workloads, as confirmed by Google's Chief Scientist Jeff Dean. The report highlights Google's unique position as a major AI model developer also producing accelerator chips at scale, attracting significant deals with companies like Meta and Anthropic. The segment also covers Blue Origin's successful New Glenn booster reuse but a payload failure for AST SpaceMobile, Cerebras's renewed IPO attempt, and the broader tech market's resilience amidst geopolitical tensions and supply chain concerns, particularly for AI infrastructure.

Key takeaway

For CTOs and AI architects evaluating future infrastructure, Google's move to inference-specific TPUs signals a critical shift in AI hardware strategy. You should assess your organization's inference demands and consider how specialized chips could optimize performance and cost, potentially exploring Google Cloud's offerings or similar specialized hardware from other vendors to secure supply and efficiency.

Key insights

Google is specializing its TPU program for AI inference to meet surging demand and maintain its competitive edge.

Principles

• Specialization enhances AI chip performance.
• Supply chain diversification is critical for AI hardware.
• First-party data improves chip design.

Method

Google leverages internal AI model teams' data to refine TPU designs, prioritizing top-tier customers and exploring specialized inference chips to optimize for growing AI workloads.

In practice

• Consider specialized inference chips for AI models.
• Diversify hardware suppliers beyond single vendors.
• Prioritize customers capable of maximizing new tech.

Topics

• Google TPUs
• AI Inference Chips
• Chip Supply Chain
• NVIDIA Competition
• Blue Origin Space Launch

Best for: CTO, VP of Engineering/Data, AI Architect, Investor, Director of AI/ML, Tech Journalist

Related on AIssential

• Google Cloud launches two new AI chips to compete with Nvidia — Google Cloud's new specialized TPUs enhance AI training and inference while complementing, not replacing, Nvidia's hardware.
• 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.
• Google Introduces Specialized Chip for New Wave of AI Computing — Google's new TPU 8t processor targets AI inference, optimizing query operations over model training.
• Can the Nvidia Monopoly on AI Chips Be Broken? — AI chip market diversity demands power-optimized, application-specific hardware supported by robust, user-friendly software stacks.
• 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 doesn't pay the Nvidia tax. Its new TPUs explain why. — Google's new specialized TPUs and vertical integration aim to reduce AI compute costs and enhance performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Bloomberg Tech.