Nvidia competitor Etched hits $5B valuation, $1B in sales for AI chip

2026-06-30 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Expert, extended

Summary

Nvidia AI chip competitor Etched, founded in 2022, has achieved a \$5 billion post-money valuation after closing an unannounced \$500 million funding round in December, bringing its total raised to \$800 million. The startup has also secured \$1 billion in contract orders for its "frontier inference clusters," full systems comprising custom chips, racks, and software designed for faster, cheaper, and more power-efficient AI inference. Etched's chips, successfully manufactured by TSMC, aim to address the significant bottleneck and cost center of AI inference. The company's technical approach focuses on low voltage inference, running at under half the voltage of GPUs to solve thermal throttling, and cluster scale memory with custom interconnects to dramatically reduce chip-to-chip latency and increase memory bandwidth for decode operations. This vertical integration strategy, from chip design to production, enables rapid development and deployment.

Key takeaway

For Directors of AI/ML evaluating future infrastructure, Etched's emergence signals a critical shift towards specialized inference hardware. You should assess whether your current general-purpose GPU clusters are creating bottlenecks and excessive costs for large-scale AI model deployment. Consider exploring vertically integrated, purpose-built inference solutions like Etched's that promise order-of-magnitude improvements in concurrency and tokens per watt, potentially enabling new, faster, and more accessible AI applications.

Key insights

Etched redefines AI chip design by optimizing for inference through low voltage and cluster-scale memory, enabling massive throughput.

Principles

• Specialized design beats general-purpose for specific use cases.
• Vertical integration accelerates product development and performance.
• Extreme urgency and risk-taking drive market velocity.

Method

Etched employs low voltage inference to prevent thermal throttling and uses cluster scale memory with custom interconnects to reduce chip-to-chip latency for high-bandwidth decode operations, disaggregating prefill and decode.

In practice

• Design for specific thermal constraints, not generic worst-case scenarios.
• Pre-fetch and parallelize development cycles to reduce time-to-market.
• Cultivate "legends" and "raw talent" for bimodal team strength.

Topics

• AI Chips
• AI Inference
• Low Voltage Inference
• Cluster Scale Memory
• Vertical Integration
• Semiconductor Manufacturing
• TSMC

Best for: CTO, VP of Engineering/Data, AI Architect, AI Hardware Engineer, Director of AI/ML, Investor

Related on AIssential

• Etched - Building AI Hardware to Make Inference Faster and Cheaper - [Invest Like the Best, EP.480] — Etched's architecture leverages low voltage inference and cluster scale memory to achieve significantly faster and cheaper AI model inference.
• After Nvidia’s $20B not-acqui-hire, AI chip startup Groq reportedly raising $650M — Groq is pivoting to an inference neocloud model, securing significant funding after a strategic Nvidia deal.
• With Nvidia Groq 3, the Era of AI Inference Is (Probably) Here — Specialized hardware for AI inference is critical due to its distinct low-latency computational requirements.
• AMD Soars on Blockbuster AI-Fueled Forecast — AI demand is driving significant growth and strategic shifts across semiconductor, cloud, and consumer tech sectors.
• This 11-person London startup wants to make AI 100x cheaper every year — AI inference costs can be drastically reduced through hardware and software optimization.
• A PhD pivot, a stint in VC and a rumoured Anthropic deal: why Europe is watching Walter Goodwin — AI inference demand is growing exponentially, creating a strategic opportunity for specialized chip development to challenge incumbents.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.