OpenAI Unveils Custom Chip With Broadcom

OpenAI and Broadcom on Wednesday unveiled their debut custom chip, called Jalapeño, marking the ChatGPT maker's first entry into artificial intelligence silicon.

The accelerator is showing cost savings of roughly 50% compared with typical AI graphics processing units, Broadcom Chief Executive Officer Hock Tan said in an interview.

Jalapeño was co-developed from initial design to manufacturing tape-out in just nine months, and the custom AI accelerator program represents what may be the fastest ASIC development cycle ever achieved in high-performance advanced semiconductors.

The chips will be made by Broadcom and used by OpenAI for inference, the compute-intensive process of serving its AI models to users in ChatGPT and other applications.

Why Build Custom Silicon Now?

AI labs such as OpenAI and Anthropic are struggling to obtain enough computing horsepower to run the latest, most powerful chatbots and coding apps.

OpenAI President Greg Brockman told CNBC that OpenAI "cannot get compute fast enough," and Broadcom CEO Hock Tan backed up that take, saying compute demand from the company's six customers is "simply insatiable," adding "It's just much more than we can address, and this is not just '26, not '27, we're seeing that same and even elevated demand in '28 as well."

The chip with Broadcom is an ASIC, which industry experts say is less flexible than Nvidia's GPU, but is also less expensive and can be designed for specific AI tasks.

The chip made by the team is as good as the Blackwell chips made by Nvidia or the tensor processing units designed by Alphabet's Google, Broadcom CEO Hock Tan said in an interview with Reuters.

Engineering samples of the Jalapeño chip are running ML workloads in the lab at production target frequency and power, including GPT-5.3-Codex-Spark, and early testing shows that Jalapeño will deliver performance per watt substantially better than current state-of-the-art.

A Nine-Month Sprint From Design to Silicon

That speed reflects deep software-hardware co-development with OpenAI's engineering teams, Broadcom's silicon implementation expertise, and the use of OpenAI models to accelerate parts of the design and optimization process.

It took the company's engineers roughly nine months to complete the chip design before it sent it to Taiwan's TSMC for manufacturing, in part because of using AI to speed specific aspects of the process, OpenAI said.

Canadian electronics manufacturer Celestica will build the server systems, which, like the chips will be used only by OpenAI.

Broadcom's silicon implementation and networking technologies, including Tomahawk networking silicon, help bring the platform to large-scale production.

The partnership builds on an October 2025 announcement. In October, after 18 months spent working together, OpenAI and Broadcom went public with plans to develop and deploy racks of OpenAI-designed chips starting late this year.

OpenAI and Broadcom announced a collaboration for 10 gigawatts of custom AI accelerators, with Broadcom to deploy racks of AI accelerator and network systems targeted to start in the second half of 2026, to complete by end of 2029.

The Energy Equation Behind AI Infrastructure

The move into custom silicon comes as OpenAI confronts massive energy demands. Initial deployment of the Jalapeño-based compute platform is planned by the end of 2026, and according to OpenAI and Broadcom, the roadmap targets gigawatt-scale data centers with Microsoft and other partners, expanding deployments over multiple generations in subsequent years.

Data centers currently account for approximately 1-2% of worldwide electricity consumption, translating to roughly 300-400 terawatt-hours (TWh) annually.

OpenAI's GPT-4, released in 2023, required an estimated 50 gigawatt-hours (GWh) of electricity for its training run—equivalent to the annual consumption of approximately 40,000 U.S. households or a small city's worth of power.

OpenAI today operates on just over 2 gigawatts of compute capacity, which Altman said has been enough to scale ChatGPT to where it is today, as well as develop and launch video creation service Sora and do a lot of AI research.

OpenAI has announced roughly 33 gigawatts of compute commitments over the past three weeks across partnerships with Nvidia, Oracle, AMD and Broadcom.

Broadcom's Custom Chip Boom

Broadcom has been one of the biggest beneficiaries of the generative AI boom by helping hyperscalers and frontier labs create their own custom chips for AI.

Shares of the chipmaker are up 10% so far in 2026 and have multiplied by almost sevenfold since the end of 2022.

Broadcom shares climbed about 2% following the announcement.

To make their own in-house chips, Meta Platforms, Amazon and Google have turned to the likes of Broadcom and Marvell, which provide specific design services and intellectual property that can be difficult to replicate in-house.

At the moment, however, because of the AI-related surge in memory demand, Broadcom's profit margin on the custom chips is not as high as some of the other chips it makes, such as networking switches, and AI chips require large amounts of high-bandwidth memory, which challenges Broadcom's margins on custom AI chip products, Tan said.

Broadcom has been the design partner for Google's Tensor Processing Unit (TPU) family since the early generations, and Google's latest TPU deployments—designed in partnership with Broadcom and manufactured on TSMC's 3nm process—are central to powering Gemini's training and inference workloads across Google's AI platform, which now serves over 750 million users.

What Changed This Week

OpenAI delivered its first custom chip after a nine-month development sprint with Broadcom, achieving what the companies claim is the fastest advanced semiconductor development cycle on record. The Jalapeño chip promises 50% cost savings over standard GPUs while matching the performance of Nvidia's Blackwell and Google's TPUs. Deployment begins by year-end as part of a 10-gigawatt buildout targeting Microsoft and other data center partners through 2029.

What to Watch

A detailed technical report on performance will be presented in the coming months. The initial deployment timeline of late 2026 will test whether OpenAI can scale custom silicon production fast enough to meet its compute demands. Anthropic is weighing building an AI chip of its own, sources told Reuters in April. Watch for similar announcements from other AI labs as the race to control the full hardware-software stack intensifies. TSMC manufacturing capacity allocation and high-bandwidth memory supply will be critical bottlenecks as multiple tech giants pursue custom chip strategies simultaneously.

Reporting based on coverage from OpenAI, Broadcom, CNBC, Bloomberg, Reuters, TechCrunch, June 24, 2026.