# OpenAI's chip move is much bigger than chips > Source: > Published: 2026-06-24 20:32:19+00:00 OpenAI has finally unveiled its long-awaited AI chip. On Wednesday, the company announced Jalapeño, its [first-generation "intelligence processor,"](https://openai.com/index/openai-broadcom-jalapeno-inference-chip/) developed in collaboration with Broadcom over nine months. The company said in its announcement that the chip was specifically designed around OpenAI's vision for "the future of LLM inference," and marks the first in a broader compute platform the companies are building together. The companies said the chip marks a major step in OpenAI's plan to build the full stack behind its models and products. In a statement, OpenAI president Greg Brockman said that, by designing more of the stack itself, the company can continue pushing for broader access to AI with greater efficiency. "The world is moving to a compute-powered economy,” Brockman said in the press release. “Jalapeño is part of our long-term full-stack infrastructure strategy to make compute more abundant, resulting in AI which is faster, more reliable, more affordable for people and businesses, and can be used to solve more important problems." OpenAI said it designed the chip with the ability to work with all LLMs, and was guided by its understanding of the industry's current inference needs. - Jalapeño's architecture is built to limit data movement while balancing compute, memory and networking resources. - The company said it's still measuring final performance, but in testing so far, performance per watt is "substantially better than current state-of-the-art." - Engineering samples of the chip are already running machine learning workloads, including those of GPT‑5.3‑Codex‑Spark, the company said. OpenAI said that Jalapeño is part of the company's "full-stack advantage," developing not only models and products on top of them, but the infrastructure underneath. This is about strengthening its progress towards better compute efficiency, better training and more powerful models. By owning the entirety of the stack, each layer is designed for the same goal, it said: "making its models faster, more reliable, and more affordable for users." The chip comes after more than a year of snags in its mission to build its own hardware. The company had initially planned to build a network of chip fabs, but [abandoned those plans in 2024](https://www.tomshardware.com/tech-industry/artificial-intelligence/openai-reportedly-builds-custom-ai-chips-as-it-embraces-amd-company-also-abandons-plans-to-build-its-own-fabs). The company then announced its partnership with [Broadcom to make chips in October](https://openai.com/index/openai-and-broadcom-announce-strategic-collaboration/) last year, ambitiously aiming to deploy 10 gigawatts of custom AI accelerators. Though this isn't the first time a major tech company has announced a custom AI chip, OpenAI's move has a few implications, Jeremy Roberts, senior director of research and content at Info-Tech Research Group, told The Deep View. "This move addresses the challenge of power consumption and efficiency of existing inference chips, reliance on Nvidia for AI hardware, and a potential entry into the datacenter hardware market, as OpenAI says that it will be compatible with other LLMs," said Roberts. ## Our Deeper *View* The Jalapeño chip is just the latest in OpenAI's quest to sink its teeth into a larger cross-section of AI's so-called [five-layer cake](https://www.thedeepview.com/articles/why-ai-may-be-more-like-electricity-than-software). In May, the company started offering [Guaranteed Capacity](https://www.thedeepview.com/articles/openai-targets-enterprise-with-guaranteed-compute), giving customers “long-term access” to its compute. It also debuted [Multipath Reliable Connection](https://www.thedeepview.com/articles/openai-targets-enterprise-with-guaranteed-compute), an open source standard for GPU networking. Each of these initiatives target layers that have nothing to do with its core product offering. And it makes sense: OpenAI wants to be foundational to every part of the industry it believes will completely transform society, consistently preaching its goal of democratizing intelligence. And while leaders at [OpenAI](https://www.thedeepview.com/articles/altman-reframes-who-controls-ai-s-future) and [other major tech firms](https://www.thedeepview.com/articles/why-ai-s-decentralization-movement-has-arrived) have started to raise concerns about the centralization of power in the industry, by embedding itself within every level of the AI stack, it is inherently making itself a cornerstone upon which all AI innovation rests.