- Anthropic reportedly considering designing its own AI chips.
- This move reflects surging demand for AI computing power.
- Company currently relies on Google and Amazon for chips.
Artificial intelligence firm Anthropic is weighing plans to design its own AI chips, signalling a potential strategic shift as the global race for computing power intensifies.
According to a Reuters report, the San Francisco-based company is in early discussions about developing in-house semiconductor capabilities, although no final decision has been made yet.
The move comes at a time when demand for high-performance chips has surged, driven by the rapid expansion of advanced AI systems and generative models.
Early-Stage Plans, No Final Commitment Yet
Sources familiar with the matter told Reuters that Anthropic is still in the exploratory phase and may ultimately choose not to pursue chip design. The company has not yet finalised a specific architecture or assembled a dedicated team for the project.
This suggests that while the idea is under consideration, it remains a long-term strategic option rather than an immediate operational shift.
AI Boom Driving Chip Demand
The discussions reflect a broader trend across the artificial intelligence industry, where access to advanced computing infrastructure has become a critical bottleneck.
Anthropic’s flagship AI model, Claude, has seen a sharp rise in demand. The company recently indicated that its annualised revenue run-rate has crossed $30 billion in 2026, a significant jump from approximately $9 billion at the end of 2025, highlighting the rapid pace of growth in the sector.
Such expansion requires massive computational resources, placing pressure on existing chip supply chains.
Reliance On Big Tech Infrastructure
At present, Anthropic depends on a mix of third-party chips to train and run its AI systems. These include tensor processing units (TPUs) developed by Google, as well as chips supplied by Amazon.
Earlier this week, the company entered into a long-term agreement with Google and Broadcom, which are involved in designing and supplying TPUs. The partnership is part of a broader push to strengthen computing infrastructure, with commitments running into tens of billions of dollars.
This reliance underscores the strategic importance of hardware partnerships, even as companies consider building internal capabilities.
Industry-Wide Shift Towards Custom Chips
Anthropic is not alone in exploring this path. Several major technology companies, including Meta and OpenAI, are evaluating or actively developing their own AI chips to reduce dependence on external suppliers and optimise performance for specific workloads.
Custom-designed chips can offer advantages such as improved efficiency, lower long-term costs and better alignment with proprietary AI models. However, the process is complex, resource-intensive and fraught with technical challenges.
High Costs And Technical Barriers
Designing advanced AI chips is an expensive undertaking. Industry estimates suggest that developing a single high-end chip can cost around $500 million.
The process requires highly specialised engineering talent, extensive testing and coordination with semiconductor manufacturers to ensure defect-free production.
These barriers mean that only well-funded companies with long-term strategic intent are able to pursue such initiatives.
The exploration of in-house chip development highlights a key shift in the AI industry: control over computing infrastructure is becoming as important as advances in software.
As competition intensifies and demand for AI capabilities continues to grow, companies are increasingly looking to secure their supply chains and build vertically integrated ecosystems.
