①Anthropic demonstrated to investors before its IPO its ability to accelerate chip R&D using Claude; ②Once AI-driven chip design tools take off, they could pose a structural threat to the existing EDA market; ③Anthropic's in-house chips may complicate its relationships with major investors such as Amazon and NVIDIA.
Cailian Press, November 4 (Editor: Li Ying) — Anthropic recently demonstrated to investors its ability to accelerate chip R&D using Claude, sparking market concerns about the impact on the traditional electronic design automation (EDA) industry; in-house chip development could also complicate Anthropic's relationships with major investors such as Amazon and NVIDIA.
According to media reports on November 3 local time, Anthropic recently emphasized its chip-design capabilities in meetings with potential IPO investors, positioning AI-assisted semiconductor engineering as a key competitive advantage for the company.
According to the report, the company's valuation in this IPO could approach $1.5 trillion.
According to media reports, Anthropic disclosed during the meeting that its newly formed 15-member hardware team is leveraging its large model Claude to accelerate custom chip development. The company stated that this approach can shorten R&D time compared with the traditional engineering processes of established vendors such as NVIDIA and AMD.
Meanwhile, Anthropic is also working to acquire more industry data. Reportedly, to enhance Claude's expertise in chip design, the company has reached out to several EDA firms, seeking access to their proprietary chip-design datasets for model optimization.
Anthropic's move reflects the intensifying competition among leading large-model developers—cutting-edge chip computing power is expensive and development cycles are long, so companies are seeking to overcome these challenges by leveraging their own models.
Competitor OpenAI has also taken action. Sarah Friar, the CFO of the ChatGPT developer, said in September that, leveraging its internal AI system and in collaboration with Broadcom, the company completed the design of its custom chip "Jalapeño" in just nine months.
Although Anthropic has not yet indicated that it will commercialize Claude's chip-design capabilities, the emergence of AI-powered design tools could pose a structural threat to the existing EDA market.
However, media reports indicate that even if Claude moves toward commercialization in the future, its adoption among third-party semiconductor companies may face resistance. Chipmakers tightly guard their proprietary design blueprints and, to prevent intellectual property leaks, often prefer deploying open-source models within isolated internal environments rather than using cutting-edge external models.
For Anthropic itself, developing its own chips could also lead to friction with its major investors. Reports indicate that these investors have collectively injected approximately $90 billion into the company and are also its primary infrastructure and cloud service providers.
Anthropic has long relied on the aforementioned investors and other hardware partners. When it publicly confirmed for the first time in August this year that it was forming a custom chip team, a company spokesperson told the media that it would continue its "multi-chip" strategy, with hardware from Amazon, Google, NVIDIA, and AMD remaining its core pillars. Therefore, if Anthropic shifts to relying on self-developed chips, its relationships with these partners could become more complicated.
On the other hand, by leveraging its in-house models, Anthropic could potentially condense a chip development roadmap that once took years into just a few months—though whether this prospect will materialize remains to be seen.
Editor/Liam