Broadcom and OpenAI Make an AI Chip Splash with Jalapeño

Over the course of the current quarter, there have been numerous announcements across the AI landscape, from chip tie-ups to locking in power capacity, and securing networking access.

One of those announcements was between Broadcom (AVGO) and OpenAI, with the two collaborating on 10 gigawatts of custom AI accelerators that would start arriving in the second half of 2026. The two have now revealed the first fruits of that labor with Jalapeño, a new chip designed specifically to run inference for large language models.

Tucked inside the announcement were three things that caught our attention.

First, the timing for the initial shipments speak to the strong ramp Broadcom telegraphed for the back half of the year on its latest earnings call.

Second, Broadcom plans to start deploying racks of AI accelerator and network systems in H2 2026 and continue through the end of 2029. That is some very nice multi-year visibility, and it’s another nod to the comment CEO Hock Tan made back in early March about a “line of sight to achieve AI revenue from chips, just chips, in excess of $100 billion in 2027.”

Remember, OpenAI is one of the five custom AI silicon customers at Broadcom, with others including Google (GOOGL), Meta (META), Anthropic, and ByteDance. It’s also well positioned to capitalize on the demand for networking we discussed with you yesterday. That combo keeps us bullish on AVGO shares.

Third, is the following from Broadcom:

By co-developing our industry-leading silicon directly with OpenAI, we are enabling the deployment of gigawatt scale data centers with Microsoft and other partners beginning in 2026.

This speaks to the existing relationship between OpenAI and Microsoft (MSFT), but it also suggests that as that capacity comes on stream, Microsoft should begin to monetize it and help ease margin pressures. We have been patient with MSFT shares and are inclined to continue to be so given the company’s position in data center and cloud infrastructure and the growing demand AI adoption and usage is placing on it.

With MSFT below our $390 pickup point, we are watching the shares closely, but being the prudent investors we are, we’re going to first let the market digest quarterly results from Micron (MU) after today’s market close.

More Pro Portfolio:

At the time of publication, TheStreet Pro Portfolio was long AVGO, GOOGL, META, MSFT and MU.