Broadcom Beats Earnings as AI Revenue Doubles, But Margin Pressure Tests Investor Conviction

Broadcom reported fiscal Q1 revenue of $19.31 billion and guided Q2 to $22.0 billion—substantially ahead of Wall Street’s $20.56 billion consensus—but the stock’s muted reaction underscores a market grappling with the profitability equation of unprecedented AI infrastructure spending.

CEO Hock Tan delivered an Earnings and revenue beat on March 4, with adjusted EPS of $2.05 versus the $2.03 estimate, as the company’s revenue grew 29% year-over-year. AI revenue surged 106% to $8.4 billion in the quarter, more than doubling year-over-year and exceeding management’s own forecast issued in December.

For fiscal Q2, Broadcom guided to $22.0 billion in revenue against analyst expectations of $20.56 billion—a beat of nearly 7%. The company projects AI semiconductor revenue of $10.7 billion next quarter, representing 143% year-over-year growth and cementing an acceleration pattern that has intensified each quarter.

Yet Broadcom shares remain down 8% year-to-date in 2026, and the stock has fallen approximately 23% from its December record. The disconnect between record results and share price performance reflects investor anxiety over gross margin compression and whether the current pace of AI capital expenditure—now forecast at $527 billion across Hyperscalers in 2026, up from $465 billion at the start of Q3 earnings season—can sustain profitable returns.

The Margin Compression Dilemma

The market’s skepticism centers on a structural shift in Broadcom’s business model. CEO Hock Tan previously acknowledged that AI sales were weighing on profit margins, with adjusted gross margin expected at approximately 77% in fiscal Q1, down from 78% the prior quarter and 79% a year ago.

Management flagged an expected sequential decline of about 100 basis points in Q1 due to a higher mix of AI revenue, including passthrough costs in rack-level solutions, as AI Semiconductors—particularly custom chips and systems—carry lower margins than traditional products, with semiconductor gross margins around 68% compared to 93% for infrastructure software, according to 24/7 Wall Street.

This margin pressure stems from Broadcom’s pivot toward system-level AI deliveries for hyperscalers. The company now provides complete rack-scale solutions—integrating custom accelerators, networking switches, and infrastructure—rather than standalone chips. While this approach locks in multi-billion dollar contracts, it introduces lower-margin hardware components and supply chain passthrough costs that compress profitability.

Infrastructure software revenue came in at $6.8 billion with only 1% year-over-year growth, further tilting the revenue mix toward hardware. The VMware integration, intended to provide high-margin software ballast, has yet to deliver the growth acceleration investors anticipated.

Hyperscaler Spending Maintains Breakneck Pace

Despite margin concerns, demand signals from Broadcom’s primary customers—Amazon, Google, Meta, and Microsoft—indicate no slowdown in AI infrastructure investment. Amazon projects $200 billion in capex for 2026, Alphabet targets $175-185 billion, Meta plans $115-135 billion, and Microsoft is tracking toward $120 billion or more, with these five companies combined planning roughly $660-690 billion on infrastructure in 2026, according to analysis from Futurum Group.

Hyperscaler capex for the ‘big five’ is now widely forecast to exceed $600 billion in 2026, a 36% increase over 2025, with roughly 75%, or $450 billion, of that spend directly tied to AI infrastructure such as servers, GPUs, and datacenters, reported the IEEE ComSoc Technology Blog.

Broadcom’s custom accelerator strategy positions it to capture substantial share of this spending. The company’s growth is primarily fueled by its leadership in hyperscaler XPU programs, spanning Google’s TPU, Meta’s MTIA, and others, with Broadcom currently having at least five identified XPU customers: Google, Meta, OpenAI, Arm/SoftBank, and ByteDance, according to research from Futurum Group.

Competitive Dynamics Intensify

Broadcom’s custom silicon approach faces mounting competition from multiple vectors. Hyperscalers such as Google, Amazon, Microsoft, and Meta are rapidly ramping proprietary accelerators like TPUs and other custom chips to reduce dependence on Nvidia, while AMD is winning new server deals—threatening margins and share over the medium term, according to analysis from Barchart.

Meta recently inked a multiyear deal with AMD involving deployment of up to 6 gigawatts of GPUs for AI Data Centers, with early shipments of MI450 GPUs in AMD’s Helios rack-scale servers beginning later this year, CNBC reported. This diversification strategy among hyperscalers introduces execution risk for Broadcom’s concentrated customer base.

Analyst Srini Pajjuri from RBC Capital expressed concern about Broadcom’s relationship with AI developer Anthropic, writing that while the supply arrangement will continue to bring in revenue, demand could fall off after the first half of 2027, according to Nasdaq.

The competitive threat extends beyond chip designers to foundry capacity constraints. Taiwan Semiconductor Manufacturing captures nearly all of the highest-end chip production work, is the largest chip manufacturer by revenue in the world, and is a leader in delivering the most advanced chips, with Broadcom, AMD, and Nvidia often working with Taiwan Semiconductor for their cutting-edge chips, noted The Motley Fool.

Market Rotation Tests AI Infrastructure Thesis

Broadcom’s stock performance mirrors broader sentiment shifts in AI infrastructure names. Nvidia beat Wall Street’s expectations and raised forward guidance due to strong demand and heavy capital expenditures planned by hyperscalers, yet the stock plunged 9.4% over the next two sessions, its worst two-day decline since April, Bloomberg reported.

The selloff is part of investors’ broader rotation away from the largest technology companies due to fears about the sustainability of the hundreds of billions of dollars committed to developing artificial intelligence capabilities. This “show me” posture from institutional investors demands evidence that current infrastructure spending will translate into durable revenue growth and acceptable returns on invested capital.

With Broadcom’s stock up nearly 60% on AI momentum over the past year, any sign of a guidance reset or margin compression could trigger a sharp ‘sell the news’ reaction, as the market has already bought the rumor and will be watching closely for reality to match or exceed it, according to analysis from AInvest.

Analyst price targets reflect this uncertainty. The average price target of $433.87 implies 31% upside from recent levels, but the spread—from a low of $300 to a high of $510—shows significant disagreement.

What to Watch

Three factors will determine whether Broadcom can convert revenue momentum into sustained share price appreciation. First, gross margin stabilization: investors need evidence that the mix shift toward AI hardware has reached equilibrium and that efficiency gains from volume production can offset lower per-unit margins.

Second, VMware monetization: the infrastructure software segment must demonstrate accelerating growth to provide the high-margin ballast that justifies the company’s premium valuation. UBS analysts point to a looming challenge with a massive wave of three-year legacy VMware contracts set to expire in late 2026 and 2027, with Broadcom’s ability to migrate these customers to new subscription-based models without significant churn essential to maintaining cash flow necessary to fund capital-intensive AI R&D, according to FinancialContent.

Third, backlog conversion velocity: with total AI order backlog at $73.0 billion across XPUs and networking components, with deliveries expected over the next 18 months, the pace at which Broadcom converts orders into recognized revenue will signal whether supply chain constraints or customer deployment timelines represent the binding constraint.

The broader question transcends Broadcom: can the semiconductor industry sustain 30-40% annual growth rates while absorbing margin compression from increasingly commoditized AI hardware? Or does the current setup—record spending, falling margins, concentrated customers—presage a profitability reckoning once infrastructure buildout peaks? Broadcom’s next several quarters will provide critical data points for that debate.