Broadcom Q3 2026 earnings beat on $16.7B in AI chips, yet light revenue guidance triggered an AVGO stock dip. Review our institutional data breakdown. (150 chars)
Broadcom Inc. (Nasdaq: AVGO) reported its fiscal third-quarter 2026 results on Wednesday, September 2, 2026, after market close, delivering revenue and adjusted earnings that beat Wall Street estimates by a comfortable margin. Yet the stock fell roughly 2.7% in the following session not because the quarter was weak, but because fourth-quarter revenue guidance came in slightly below what analysts had modeled. That gap between an unambiguously strong Broadcom Q3 2026 earnings report and a guidance figure that fell just short of consensus is the central tension investors are now trying to interpret.
This article examines what actually happened in the quarter, why AI semiconductor demand has become the dominant force in Broadcom's business, what the numbers reveal about the parts of the company that are not benefiting from the AI buildout, and the risks and uncertainties that complicate an otherwise bullish narrative.
AI Semiconductor Revenue More Than Tripled, Eclipsing the Rest of the Business
The headline figures were strong across the board. Broadcom generated $29.59 billion in consolidated revenue for the quarter ended August 2, 2026, up 86% from $15.95 billion a year earlier, and above the roughly $29.25–$29.36 billion analysts had expected. Non-GAAP (adjusted) earnings per share came in at $3.32, beating the consensus range of about $3.21–$3.24 by roughly 3%.
On a GAAP basis, net income more than tripled year-over-year to $13.09 billion, or $2.68 per diluted share, up from $4.14 billion, or $0.85 per share, a year earlier. It is worth being precise here: the $2.68 GAAP figure and the $3.32 non-GAAP figure are both accurate numbers from the same earnings release, but they measure different things. Non-GAAP EPS strips out items such as stock-based compensation and acquisition-related amortization, which is why it is consistently higher than the GAAP figure a distinction that matters for anyone comparing Broadcom's results to analyst estimates, since Wall Street consensus is typically built around the non-GAAP number.
The engine behind these results was unmistakable. AI semiconductor revenue reached $16.7 billion, growing 221% year-over-year and 54% quarter-over-quarter, comfortably ahead of the roughly $15.2 billion analysts had penciled in. On the earnings call, Chief Executive Hock Tan attributed the surge to strong demand from six XPU (custom AI accelerator) customers, which the company has previously identified as including Google, Anthropic, and OpenAI. This "six customers" framing is significant because it is the specific detail most financial coverage has centered on it tells readers not just how large AI revenue has become, but how concentrated the customer base generating that revenue actually is. BroadcomYahoo Finance
Profitability metrics moved in tandem with revenue. Operating margin hit a record of roughly 68% of revenue, and free cash flow reached a record $13.7 billion, equal to about 46% of revenue. Both figures reflect the operating leverage inherent in custom-chip manufacturing at scale: once design and production costs are amortized across large-volume orders, incremental revenue converts to profit and cash at a high rate.
Non-AI Chips and Software Are Growing at a Fraction of the AI Segment's Pace
The quarter's most important nuance is not the AI headline number itself, but the gap between AI and everything else Broadcom sells. Non-AI semiconductor revenue the company's traditional chip business serving broadband, wireless, storage, and industrial customers totaled just $4.2 billion, up only 5% year-over-year and essentially flat sequentially. Infrastructure software revenue, largely driven by the VMware business Broadcom acquired in 2023, came in at $8.75 billion, slightly below the $8.82 billion analysts had expected.
| Segment | Q3 FY2026 Revenue | YoY Growth | Vs. Consensus |
| AI semiconductor | $16.7 billion | 2.21 | Beat (~$15.2B est.) |
| Non-AI semiconductor | $4.2 billion | 0.05 | Flat sequentially |
| Infrastructure software | $8.75 billion | +29% YoY | Slight miss (~$8.82B est.) |
| Total consolidated | $29.59 billion | 0.86 | Beat (~$29.25–29.36B est.) |
This table illustrates why analysts increasingly describe Broadcom less as a diversified semiconductor company and more as a company with one extraordinarily fast-growing AI division layered on top of a mature, slower-growing base. The mechanism matters for how the market values the stock going forward: if AI semiconductor growth decelerates even modestly, the underlying non-AI business does not have comparable momentum to offset it, which increases the company's sensitivity to swings in AI infrastructure spending by its hyperscale customers.
A Slightly Light Guide, Not a Weak Quarter, Explains the Stock's Pullback
If the quarter itself was strong, why did AVGO shares fall? The answer lies in guidance rather than results. For the fiscal fourth quarter, Broadcom said it expects revenue of $34.8 billion, while analysts were expecting $35.03 billion, according to LSEG. That guidance still implies roughly 93% year-over-year growth an extraordinary pace by any conventional standard but it sits modestly below what the Street had already built into its models. CNBC
This is a recurring pattern in high-growth technology earnings: when a stock has run up sharply on the expectation of continued acceleration, even a minor shortfall relative to the most optimistic forecasts can trigger a sell-the-news reaction, regardless of how strong the underlying quarter was. Shares declined approximately 2.7% in the session following the release, and as of September 11, 2026, AVGO stock was consolidating in a range of roughly $353.70 to $360.16, with a market capitalization near $1.70 trillion. It is worth noting that on the day of the report itself, the company's market value had briefly touched closer to $1.8 trillion before the guidance-driven pullback took hold, according to CNBC's coverage of the results. The available evidence points to the guidance-versus-expectations gap, not any deterioration in the quarter's actual performance, as the specific mechanism behind the share-price move.
Management also used the call to extend its forward outlook well beyond the immediate quarter. The company raised its fiscal 2026 AI revenue guidance to $58 billion, up 186% year-over-year, and provided targets for AI semiconductor revenue to roughly double to $115 billion in fiscal 2027 and $230 billion in fiscal 2028. Multi-year guidance of this specificity is unusual, and some analysts have linked the move to a broader pattern in which large semiconductor suppliers are trying to reassure markets that remain uncertain about whether AI infrastructure demand is durable or approaching an "air pocket", following a similar multi-year guide from a major competitor days earlier. Yahoo FinanceLighthouse-canton
Six Customers Now Anchor a Business Worth Tens of Billions of Dollars
The customer-concentration dynamic deserves closer examination because it is both the source of Broadcom's growth and its most cited risk. Custom AI accelerators, often called XPUs, are chips designed to a specific customer's specifications for training or running large AI models an alternative to buying general-purpose GPUs from merchant suppliers. Broadcom's role is to co-design and manufacture these chips for hyperscale customers who want AI compute tailored to their own workloads and, in some cases, want to reduce dependence on a single GPU supplier.
Reporting on the earnings call indicates that Anthropic is expected to become one of Broadcom's largest XPU customers by fiscal 2027, with a disclosed deployment schedule that includes gigawatt-scale chip volumes ramping through 2027 and 2028. The mechanism connecting this disclosure to Broadcom's financial outlook is straightforward: as customers commit to multi-year, multi-gigawatt chip programs, Broadcom gains revenue visibility it can translate into forward guidance which is part of why management felt confident enough to guide two fiscal years ahead rather than the usual single quarter. Lighthouse-canton
That said, concentration among a handful of customers means Broadcom's AI segment performance is tied to the capital-spending decisions of a small number of very large technology companies. If any of these customers were to slow, delay, or restructure their AI infrastructure investment plans, the effect on Broadcom's growth rate would be more pronounced than it would be for a supplier with a broader customer base.
Data-Center Capacity Constraints Could Cap How Fast Broadcom's Backlog Converts to Revenue
Beyond customer concentration, management identified a more physical constraint on growth during the earnings call. Executives confirmed that the company's long-term outlook is highly sensitive to land, power, and data-center shell availability, noting that construction projects for AI data centers carry long lead times and that this factor is directly reflected in the conservative multi-year forecasts provided for fiscal 2027 and 2028.
This is a meaningful, non-sensational risk: even if customer demand for Broadcom's chips remains robust, the pace at which that demand converts into shipped, revenue-generating volume depends on how quickly hyperscale customers can bring new data-center capacity online a process governed by electricity availability, permitting, and construction timelines that are largely outside Broadcom's control. It represents an emerging, acknowledged risk rather than a confirmed problem, since management explicitly framed its 2027–2028 guidance as already accounting for these constraints.
Broadcom's Software Business Grew Steadily But Missed a Narrow Estimate
While AI semiconductors dominated headlines, the infrastructure software segment's performance is a useful reminder that not every part of a diversified technology company moves at the same speed. Infrastructure software revenue grew 29% year-over-year to $8.75 billion a healthy expansion rate on its own, driven in large part by continued enterprise adoption of VMware Cloud Foundation but it landed marginally below the roughly $8.82 billion consensus estimate. The shortfall was small in absolute terms, but it underscores that Broadcom's non-AI businesses, while growing, are not generating the kind of upside surprises that have become routine in the AI semiconductor segment.
Risks, Headwinds, and Identified Problems Investors Should Track
Several distinct categories of risk merit attention, each with a different level of certainty attached:
- Confirmed risk customer concentration. Broadcom's AI revenue growth is tied to a small, named group of six XPU customers. A change in spending plans from any one of them would have an outsized effect relative to a more diversified customer base.
- Confirmed risk guidance sensitivity. The stock's reaction to a modest guidance shortfall demonstrates that AVGO shares are priced for continued acceleration; any future guide that falls short of consensus, even while showing strong absolute growth, could trigger similar volatility.
- Emerging risk physical infrastructure constraints. Management's own disclosure that land, power, and data-center shell availability could limit how quickly demand converts to revenue is a real, acknowledged constraint on the pace of growth, not a hypothetical one.
- Scenario risk AI capex cycle durability. The available evidence does not indicate that AI infrastructure spending is slowing, but the market's own skepticism reflected in commentary tied to Broadcom's and a competitor's decision to issue unusually long-dated guidance shows that questions about the durability of this capital-spending cycle remain unresolved.
- Non-AI segment stagnation. With non-AI semiconductor revenue growing just 5% year-over-year and roughly flat sequentially, that portion of the business is not providing meaningful diversification against a slowdown in AI demand.
None of these risks currently point to deterioration in Broadcom's reported results. Rather, they identify the conditions under which the company's growth trajectory and the market's valuation of that growth could be tested in future quarters.
A Thematic, Not Direct, Link to the AI Capex Story Reaching Other Markets
Broadcom's results are also relevant to a broader theme playing out across markets: the same AI capital-expenditure cycle fueling demand for Broadcom's custom chips has prompted some Bitcoin mining companies to pivot toward hosting AI and high-performance computing workloads, seeking to redeploy power infrastructure originally built for crypto mining. This is a thematic parallel rather than a direct connection Broadcom's earnings release does not address cryptocurrency markets, and there is no indication in the company's disclosures of any direct relationship between its AI chip business and crypto-mining infrastructure. Any speculation linking the two should be understood as commentary on parallel capital flows into AI infrastructure, not as a claim about Broadcom's own business strategy.
What to Watch Before Broadcom's Next Report
Broadcom is scheduled to report fiscal fourth-quarter and full fiscal-year 2026 results after market close on Wednesday, December 9, 2026. Between now and then, several data points will help determine whether the current narrative accelerating AI demand offset by a maturing non-AI business and periodic guidance-driven volatility continues to hold. These include whether Q4 AI semiconductor revenue accelerates as management indicated on the call, whether infrastructure software growth stabilizes or reaccelerates, and whether commentary from Broadcom's named hyperscale customers about their own AI capital spending remains consistent with the multi-year volume commitments Broadcom has disclosed.
The Broader Takeaway From the Broadcom Q3 2026 Earnings Report
Taken together, the evidence from this Broadcom Q3 2026 earnings report supports a fairly specific conclusion: Broadcom's near-term financial performance is now overwhelmingly a function of custom AI chip demand from a handful of very large customers, and its stock's short-term price behavior is highly sensitive to whether forward guidance meets, exceeds, or merely approaches consensus expectations. The quarter itself showed genuine strength record revenue, record margins, record free cash flow while the post-earnings share-price decline reflected a guidance nuance rather than any weakness in actual results. Readers evaluating this report should weigh the strength of the AI segment's growth against the concentration and infrastructure-capacity risks management itself has flagged, rather than treating either the beat or the guidance miss in isolation as the full story.