Anthropic is moving closer to the custom chip conversation that now surrounds many large AI companies. After Reuters reported back in April that the company was considering making its own AI chips in response to chip shortages, The Information reported on Thursday that Anthropic had been in contact with Samsung about a possible collaboration.
The talks still appear early. According to the report, Anthropic has not decided what the chip would be used for, how it would fit into the server, or how powerful it would be. That leaves the most important technical and strategic questions unresolved, even as the direction of travel is clear: AI companies are looking more seriously at hardware built for their own needs.
What Anthropic is reportedly discussing
The reported Samsung discussion centers on a pending chip. The source article does not say that Anthropic has committed to building it, nor does it describe a finished design. Instead, the key point is that the company is exploring whether Samsung could be part of a custom chip effort.
That distinction matters. A custom AI chip can mean many things depending on the workload, the server design, and the performance target. The report specifically says Anthropic has not yet decided what the chip will be used for, how it will fit into the server, or how powerful it will be.
In other words, the work is not being described as a completed product plan. It is better understood as a potential hardware path that Anthropic is evaluating while it continues to rely on existing partners for compute.
Anthropic says diversified hardware remains central
When TechCrunch asked Anthropic for comment, the company pointed to the importance of a diversified hardware stack. Anthropic said chips from Google, Amazon, and Nvidia will continue to be pivotal to its compute strategy.
On the potential Samsung partnership, Anthropic said it had nothing further to add. That response keeps the company from confirming any specific Samsung plan while still making clear that compute strategy is a major focus.
The wording also shows that a custom chip would not necessarily replace existing suppliers. Based on Anthropic’s comment, the company still sees Google, Amazon, and Nvidia as important to its hardware mix. A Samsung-linked chip, if it happens, would need to fit into that broader stack rather than stand alone as the whole strategy.
Why AI companies are looking beyond standard chips
The broader industry context is straightforward: AI companies need enormous amounts of compute, and a number of them have sought to develop custom chips. The source article gives two reasons for that interest.
- Specific compute tasks: custom hardware can be designed for particular AI workloads rather than used as a general solution for every case.
- Independence from Nvidia: companies may want some distance from a supplier that remains the undisputed leader of the chip industry.
That does not mean Nvidia is being pushed aside. In Anthropic’s own comment to TechCrunch, Nvidia remains part of the company’s compute strategy. The point is that major AI companies appear to be looking for more control, more variety, and more options in their hardware pipelines.
Chip shortages are another part of the backdrop. Reuters reported back in April that Anthropic was considering producing its own AI chips as a way to respond to shortages. If supply is a constraint, custom chip work can become not only a performance question but also a planning question.
OpenAI, Amazon, and Google frame the race
Anthropic is not operating in isolation. The source article notes that the company’s move may also be a response to an announcement last week from OpenAI, its key competitor. OpenAI has teamed up with Broadcom to announce its own custom-built inference processor, called “Jalapeño.”
OpenAI says that chip is more efficient and shows better performance-per-watt than competitor chips. The source does not provide more technical detail, but the comparison is still significant: AI companies are increasingly discussing hardware in terms of efficiency, workload fit, and competitive position.
Cloud providers are also part of this landscape. Amazon and Google both offer custom-built TPUs as part of their cloud offering. Anthropic already referenced Google and Amazon in its comment about a diversified hardware stack, which underscores how closely AI development is tied to cloud infrastructure and specialized compute.
Why Samsung is a logical name in the mix
Samsung is already deeply connected to the AI chip industry. The source article says Samsung is a major partner of Nvidia and produces chips that Nvidia needs to train or run its AI models. In turn, Samsung uses Nvidia’s software to manufacture its chips.
Samsung and Nvidia are also working on an AI chip factory in South Korea. Separately, Samsung has discussed partnering with Google on chip-making efforts. Those relationships put Samsung near several important parts of the AI hardware supply chain.
For Anthropic, that makes Samsung a notable potential collaborator, even if no final plan has been announced. The company is reportedly exploring a custom AI chip at a time when rivals and infrastructure providers are already investing in specialized processors. The unresolved details are still large, but the strategic direction is easy to read: compute has become one of the central battlegrounds in AI.