Beyond AI models: Anthropic is exploring a custom AI chip with Samsung. Here's why it matters
&imwidth=600&imheight=450&format=webp&quality=medium "Beyond AI models: Anthropic is exploring a custom AI chip with Samsung. Here's why it matters")
Beyond AI models: Anthropic is exploring a custom AI chip with Samsung Photograph: (X)
Anthropic is reportedly discussing a potential partnership with Samsung to develop a custom AI chip.
The artificial intelligence race is no longer just about building smarter AI models. It is increasingly becoming a battle over the chips that power them. Anthropic, the company behind Claude AI, is reportedly in discussions with Samsung about developing a custom AI chip, according to a report by The Information. While the talks are still at an early stage, the move suggests Anthropic is exploring ways to strengthen its long-term computing strategy as demand for AI hardware continues to grow. The company has not confirmed a partnership. In a statement to TechCrunch, Anthropic said chips from Nvidia, Google and Amazon will continue to play a key role in its AI infrastructure. It declined to comment further on reports about Samsung.
Why does Anthropic want its own AI chip?
Training and running advanced AI models requires enormous computing power. Today, much of that work depends on Nvidia's AI chips, which remain the industry standard. However, growing demand has led to supply constraints and rising costs, prompting several technology companies to develop their own processors. A custom chip could allow Anthropic to optimise hardware specifically for Claude AI, improve energy efficiency, reduce operating costs and gain greater control over its computing infrastructure. According to Reuters, Anthropic had already begun exploring the idea of designing its own AI chips earlier this year.
Samsung could become an important partner
Samsung is one of the world's largest semiconductor manufacturers and already works closely with Nvidia on AI hardware. The South Korean company also manufactures advanced chips for other technology firms and has expanded investments in AI chip production. A partnership with Anthropic could strengthen Samsung's position in the rapidly growing AI semiconductor market. However, several important decisions reportedly remain unresolved, including what the chip will be used for, how powerful it will be and how it will fit into Anthropic's server infrastructure.
AI companies are racing to control their hardware
Trending Stories
&im=FitAndFill=(700,400))
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
)
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
Anthropic is not alone.
OpenAI recently announced its custom inference chip, Jalapeño, developed with Broadcom. Google has long relied on its in-house Tensor Processing Units (TPUs), while Amazon designs its own Trainium and Inferentia chips for Amazon Web Services.
The shift reflects a broader trend across the AI industry. Instead of relying entirely on Nvidia, companies are building specialised chips that better match their own software and workloads. For Anthropic, a custom processor would represent more than a hardware upgrade. It could become an important step towards reducing dependence on third-party suppliers while preparing for increasingly powerful AI systems in the years ahead.
