Broadcom's AI momentum meets margin math: what finance teams should watch
Broadcom shares dropped more than 11% after issuing a gross margin outlook that signaled pressure, even as results beat expectations across the board. The company guided to a 76.9% gross margin for the current quarter, down from 79.0% a year ago and below 77.9% in Q4 and 78.3% in Q3. That step-down erased over $200 billion in market value in one session.
The paradox: demand is red-hot. AI chip revenue jumped 74% to $6.4 billion (vs. ~$6.2 billion expected), overall revenue rose nearly 30% to $18 billion (vs. ~$17.5 billion expected), and EPS climbed to $1.95 from $1.42. Broadcom also cited AI orders from Anthropic surging to $21 billion.
What actually moved the stock
Investors fixated on mix and pass-through dynamics. Reports pointed to an additional $11 billion of Anthropic orders tied to server racks built around Google's TPU chips, where Broadcom passes along costs with little to no markup. That lifts revenue but compresses gross margin, at least near term.
Analysts summed it up bluntly: AI revenue is scaling fast, but the cost of growth is showing up in margins. The market wants proof that Broadcom can keep unit momentum while defending profitability as systems revenue and pass-through components become a larger share.
Context: the AI trade is being repriced
Tech sold off broadly after updates from Broadcom and Oracle, with the Nasdaq Composite down about 1.4% on Friday. Oracle fell over 14% a day earlier on higher AI-related costs. Even after the slide, Broadcom is still up roughly 3% over the past month, while Oracle is down nearly 20% and almost 50% from September highs.
For years, most semis traded like commodity businesses. AI flipped that script as a handful of suppliers, led by Nvidia, controlled scarce, specialized chips and priced accordingly. The new worry: as more capacity arrives and competition broadens, pricing power could ease and margins may drift lower toward historical norms.
Positive offsets you can't ignore
Broadcom's AI pipeline is deep. The company signed a large custom-chip agreement with OpenAI for 10 gigawatts of capacity, with revenue impact expected to start in 2027. Street commentary calling the sell-off "AI stock angst" suggests many still view the margin hit as timing and mix, not demand destruction.
If Broadcom can scale full AI systems - chips plus networking and racks - and shift more revenue into higher-margin components over time, gross margin could stabilize even as AI revenue compounds. That's the crux of the bullish case.
What finance leaders should track next
- Gross margin trajectory: Does GM hold ~76%-77% or slip further as pass-through racks scale? Watch mix commentary and pricing on custom silicon vs. systems.
- Revenue mix: Split between custom AI ASICs, networking, and pass-through content. More proprietary content should support margins.
- Contract structure: Any updates on take-or-pay, pricing floors, and cost pass-through. These determine how shocks flow through to GM and cash.
- Supply and competition: New AI chip entrants and foundry capacity ramps that could pressure pricing as the market matures.
- Customer concentration: Exposure to Google, Anthropic, and OpenAI; timing of the OpenAI program's revenue ramp.
- Cash conversion: Working capital needs for systems, inventory turns, and any changes to capital intensity.
- Capital returns: Buyback/dividend flexibility if volatility persists.
Positioning takeaways
The setup is a classic beat-and-worry: top-line and AI orders look strong, but margin math is doing the talking. If your thesis rests on Broadcom as a systems supplier, near-term GM pressure is part of the journey - and the key is whether mix and scale improve unit economics into 2026-2027.
For portfolio construction, stress-test scenarios where GM runs 75%-77% for several quarters vs. a rebound to the high-77s. Tie that to free cash flow per share and required returns on AI capacity. Use volatility to assess whether sentiment has overshot the fundamentals as the OpenAI program gets closer.
Bottom line
AI demand is intact. The debate is margins. If Broadcom proves it can convert large AI orders into durable, higher-margin systems revenue, the multiple can hold. Watch guidance, mix disclosures, and early signals from the OpenAI build starting in 2027.
If you're mapping AI tooling for finance teams and workflows, this curated list may help: AI tools for finance.
Your membership also unlocks:
