Broadcom gets major OpenAI boost in AI chip race

Broadcom (AVGO) just got a new validation point in the artificial intelligence chip race: OpenAI’s first custom chip, Jalapeño, unveiled on June 24.

The chip was designed with Broadcom and built for inference, the computing process that allows products such as ChatGPT to respond to user requests.

For Broadcom, the project is another sign that the chipmaker is becoming one of the key behind-the-scenes partners for companies building custom AI silicon around their own large-scale workloads.

Broadcom shares traded near $382 after the announcement, as investors weighed what OpenAI’s chip push could mean for the company’s AI business.

The question now is whether Jalapeño gives Broadcom another durable growth opportunity as major technology companies look beyond standard graphics processors and build chips tailored to their own AI needs.

OpenAI gives Broadcom a new custom chip role

OpenAI and Broadcom said in company statements that Jalapeño is the first step in a multigeneration compute platform with initial deployment planned by the end of 2026.

The companies said OpenAI designed the chip based on its understanding of large language models, while Broadcom is helping industrialize the platform through silicon implementation, high-performance networking, and scalable production systems.

That makes the announcement more than a one-time chip launch for Broadcom. It places the company within OpenAI’s effort to build infrastructure for its own AI workloads, rather than relying solely on standard chips available to the broader market.

Related: Broadcom CEO drops unexpected message on AI and software

Rather than serving as a general-purpose AI accelerator, Jalapeño is designed specifically for large language model inference, the everyday computing work that happens after an AI model has been trained. Every user prompt, chatbot response, and coding request can add to demand for computing capacity, making inference one of the biggest infrastructure challenges as AI products scale.

OpenAI said engineering samples of the chip are already running machine-learning workloads in its lab at the target frequency and power, including GPT-5.3-Codex-Spark.

Reuters reported that OpenAI plans to deploy Jalapeño by the end of this year, and that the chips and related server systems will be used only by OpenAI.

Broadcom’s OpenAI boost comes with one investor test

The OpenAI announcement strengthens Broadcom’s position in a market where more AI companies are designing chips tailored to their own workloads.

More AI:

• Goldman Sachs has blunt message for AI stock investors
• Microsoft CEO sends a blunt warning on AI and the tech ecosystem
• The next AI infrastructure race has nothing to do with chips

Nvidia (NVDA) still dominates the market for graphics processing units used to train and run AI models. But OpenAI’s Jalapeño project shows why some of the largest AI companies want more control over the chips, networking, and systems that power their own products.

That shift gives Broadcom a different opening in the AI chip race. Instead of selling a standard processor to every customer like Nvidia, Broadcom helps large technology companies turn their own chip designs into systems that can run at scale.

Reuters reported that Meta Platforms (META), Amazon (AMZN), and Alphabet’s (GOOGL) Google have also turned to companies such as Broadcom and Marvell Technology (MRVL) for design services and intellectual property that can be difficult to reproduce in-house.

Related: JPMorgan's latest Broadcom outlook sends key signal

However, the custom chip opportunity comes with a more complicated investor question.

Reuters reported earlier in June that Broadcom expected AI chip revenue of $16 billion in its current third quarter, slightly below analysts’ estimates of $16.36 billion, according to Visible Alpha. CEO Hock Tan also stuck with the company’s long-range forecast of $100 billion in sales from AI chips, according to Reuters.

It shows investors that Broadcom is still winning high-profile AI chip work, even after Wall Street became more demanding about AI chip growth expectations.

SOPA Images &sol Getty Images

Key questions for Broadcom investors

Investors are likely to focus on several Broadcom-related questions after the OpenAI announcement:

• When OpenAI deploys Jalapeño
• Whether the project expands into a larger multigeneration platform
• How much custom AI chip revenue can Broadcom generate
• Whether other AI customers follow OpenAI’s path
• How custom chip margins compare with Broadcom’s other semiconductor products
• Whether high-bandwidth memory costs remain a drag
• How Broadcom competes with Marvell in custom silicon

Meanwhile, margins may become one of the most important questions.

Reuters reported that Broadcom’s profit margin on custom AI chips is not as high as some of the company’s other chips, such as networking switches, because AI chips require large amounts of high-bandwidth memory.

That means the OpenAI announcement is a clear win for Broadcom’s positioning, but not a complete answer for investors.

Broadcom still has to show that custom AI chip projects can scale into large, profitable revenue streams. Those projects also depend on customer deployment timelines, memory supply, manufacturing partners, and the ability to turn specialized designs into systems that can run at scale.

Broadcom’s AI chip role gets harder to ignore

OpenAI’s Jalapeño chip does not change the AI chip market overnight. But OpenAI’s first custom chip shows that major AI companies are seeking greater control over the systems that run their models.

For now, Jalapeño gives Broadcom another high-profile AI win and a clearer place in one of the most important infrastructure battles in technology. It does not need to replace Nvidia to benefit from the race. Instead, it needs to prove that more customers want chips built around their own AI workloads, and that Broadcom can help turn those designs into scalable systems.

Related: Broadcom CEO unnerves biggest AI backers in rattling pivot