Here’s something the market has gotten a little sideways on. Broadcom just posted one of the most impressive quarters of any chip company in recent memory, and the stock is sitting roughly 25% below its 52-week high.
That disconnect is worth paying attention to.
What the Numbers Actually Show
In its fiscal second quarter of 2026 (ended May 3, 2026), Broadcom reported $22.187 billion in total revenue, up 48% year over year. The headline, though, is the AI number. AI semiconductor revenue hit $10.8 billion in Q2, growing 143% year over year, driven by demand for custom AI accelerators and AI networking. The quarter before that, Broadcom had guided to Q2 AI semiconductor revenue of about $10.7 billion (roughly 140% growth), and Q2 results came in slightly above that.
CEO Hock Tan then guided for Q3 AI semiconductor revenue of $16 billion, which would represent more than 200% growth year over year. Total Q3 revenue guidance came in at $29.4 billion, up 84% year over year.
These are not incremental numbers. This is a business scaling at a pace that very few semiconductor companies have ever managed.
Slight tangent, but worth knowing: the stock’s June selloff wasn’t really about the results. It was about expectations. Broadcom’s $22.187 billion Q2 revenue was a modest miss versus some street estimates (one widely cited consensus was about $22.27 billion), and Tan reiterated (rather than raised) the company’s longer-term AI semiconductor target. Investors who had priced in a raise sold. The business kept running.
The Customer List Is the Real Story
Broadcom has discussed “six core customers” supporting its AI semiconductor growth, but it has not publicly confirmed the exact lineup in the way this paragraph states, and not all of the names listed here are verifiable as “named custom-chip customers.” What is clear is that Broadcom’s AI momentum is being driven by multi-year programs spanning hyperscale customers and leading AI companies, including a newly public partnership with OpenAI.
OpenAI joined the public story this year. On June 24, 2026, OpenAI and Broadcom unveiled a co-developed AI chip called Jalapeño, positioned as OpenAI’s first “Intelligence Processor” focused on inference. OpenAI and Broadcom have described an initial deployment targeted by the end of 2026, with expansion in subsequent years.
Broadcom has also reiterated that it has “line of sight” to AI semiconductor revenue in excess of $100 billion in fiscal 2027.
The Apple Deal Nobody Saw Coming
Then, just this week, Broadcom and Apple announced an expanded multi-year custom silicon partnership through 2031 that Apple says is expected to exceed $30 billion. Apple said the agreement is expected to lead to the production of more than 15 billion U.S.-made chips and includes plans for Broadcom to expand and modernize manufacturing facilities in Fort Collins, Colorado.
This matters beyond just the revenue number. One of the lingering concerns heading into the selloff was whether Apple might eventually move its chip design fully in-house, cutting Broadcom out. The 2031 extension, with an expanded scope, answers that question pretty directly.
The Valuation Argument
At roughly $370 per share in early July, Broadcom traded around the low-to-mid-50s on trailing P/E (based on common market data at the time), not ~60x. The specific comparisons cited here (industry average ~75x and peer average ~81x), as well as the exact “$501 to $517” consensus target range and “26 analysts” count, could not be verified to a single reliable, consistent source and should be treated as directional rather than precise.
The forward multiple, supported by revenue growth guided to exceed 80% in the coming quarter, makes the discount look larger than the danger suggests.
What Could Go Wrong
The customer concentration is real. If a major hyperscaler scales back spending or shifts strategy, the revenue impact would be material. The $100 billion 2027 target also requires continued execution across multiple large programs, with no disruption to the order pipeline.
Erste Group downgraded AVGO to Hold earlier this week, citing valuation concerns. So this is not a story with no skeptics. It never is.
What’s interesting, though, is the combination of factors now in place: an Apple commitment expected to exceed $30 billion locked through 2031, AI semiconductor revenue growing at triple-digit rates, an OpenAI partnership just getting started, and a stock well off its highs. The next earnings report isn’t until September, which gives the stock time to either rebuild confidence or face another round of expectations management.
The question isn’t whether the AI custom chip market is growing. It clearly is. The question is whether Broadcom is the right company to own inside that market. Its customer list may be the strongest argument anyone can make.
