Broadcom delivered one of the strongest quarters in its history on June 3, 2026 – and then watched its stock fall by as much as 13% in the days that followed. The chipmaker reported fiscal Q2 2026 revenue of $22.2 billion, up 48% year over year, driven by a record $10.8 billion in AI semiconductor sales that grew 143% from a year earlier. By almost every operating metric, it was a blowout. Yet the market reaction exposed a deeper anxiety running through the entire AI trade in mid-2026: investors are no longer asking whether AI chip demand is real. They are asking how much future growth is already priced into the stocks.

This is the paradox at the center of the latest Broadcom earnings report. The company beat expectations, raised guidance to levels that would have seemed implausible two years ago, and projected $56 billion in full-year AI semiconductor revenue. And the reward was a double-digit selloff that dragged much of the AI hardware complex lower alongside it. Below is a full breakdown of the numbers, the customer dynamics behind Broadcom’s custom-chip empire, how the business stacks up against Nvidia, and what the reaction tells us about where the AI build-out goes next.

Broadcom Q2 2026 Earnings: The Headline Numbers

Broadcom announced its fiscal second-quarter 2026 results after the close on Wednesday, June 3, 2026, covering the quarter that ended May 3, 2026. The top line came in at $22.187 billion, a 48% increase year over year and ahead of the company’s own guidance. Non-GAAP diluted earnings per share were $2.44, topping the consensus estimate of $2.40 by four cents, according to MarketBeat. On a GAAP basis, diluted EPS was $1.91.

Profitability was the real story. Broadcom posted GAAP net income of $9.310 billion and non-GAAP net income of $12.074 billion for the quarter. Adjusted EBITDA reached $15.244 billion, equal to 69% of revenue – a margin profile that very few companies at this revenue scale can match. Cash generation was equally striking: cash from operations hit $10.493 billion, and free cash flow came in at $10.262 billion, or 46% of revenue. The company also declared a quarterly dividend payable on June 30, 2026 to stockholders of record as of June 22, 2026.

Those are not the numbers of a company in trouble. They are the numbers of a business converting nearly half of every dollar of revenue into free cash flow while growing the top line at close to 50%. The disconnect between that performance and the share-price reaction is precisely what made this earnings event a defining moment for the 2026 AI narrative.

The $10.8 Billion AI Semiconductor Engine

Broadcom’s semiconductor solutions segment generated a record $15.0 billion in revenue during the quarter, up 79% year over year. Inside that figure sits the number Wall Street fixates on every quarter: AI semiconductor revenue, which reached a record $10.8 billion and grew 143% from the prior-year period. AI is no longer a slice of Broadcom’s chip business – it is the majority of it.

CEO Hock Tan framed the quarter directly on the earnings call. “In our fiscal Q2 2026, total revenue reached a record $22.2 billion, up 48% year on year, above our guidance on strength in AI semiconductors,” he said. On the AI line specifically, Tan added: “Driving this growth was AI semiconductor revenue at a record at 10.8 billion, up 143% year on year.” On a trailing-twelve-month basis, Broadcom’s semiconductor solutions sat at roughly $48 billion, with commentary around the report attributing about three-quarters of that to AI.

What separates Broadcom’s AI business from the rest of the field is its composition. This is not primarily a merchant GPU business. It is built on two pillars: custom AI accelerators – the XPUs and ASICs that hyperscalers design in partnership with Broadcom – and the high-speed networking silicon (Tomahawk and Jericho switches, plus optical and PCIe connectivity) that ties those accelerators together into massive training clusters. As model training scales to tens of thousands of chips, the networking fabric becomes as critical as the compute, and Broadcom sells both halves of that equation.

Broadcom Q2 FY2026 Financial Scorecard

The table below summarizes the key reported metrics for the quarter ended May 3, 2026, alongside their year-over-year movement where disclosed.

Metric	Q2 FY2026	YoY / Note
Total revenue	$22.187B	+48%
Semiconductor revenue	$15.0B (record)	+79%
AI semiconductor revenue	$10.8B (record)	+143%
Infrastructure software revenue	$7.2B	+9%
GAAP net income	$9.310B	–
Non-GAAP net income	$12.074B	–
GAAP diluted EPS	$1.91	–
Non-GAAP diluted EPS	$2.44	Beat consensus $2.40
Adjusted EBITDA	$15.244B	69% of revenue
Free cash flow	$10.262B	46% of revenue

Guidance: $16 Billion in AI Chips Next Quarter

If the reported quarter was strong, the outlook was extraordinary. Broadcom guided to fiscal Q3 2026 revenue of approximately $29.4 billion, which the company said would represent 84% year-over-year growth. Management paired that with margin guidance of roughly 67% non-GAAP operating income margin and about 68% adjusted EBITDA margin – meaning the company expects to scale revenue by tens of billions without sacrificing its industry-leading profitability.

On the AI line, Tan guided to $16 billion in AI semiconductor revenue for fiscal Q3, which he said would be up more than 200% year over year. He went further, projecting roughly $56 billion in AI semiconductor revenue for full-year fiscal 2026 – an increase of approximately 180% over fiscal 2025. To put that in perspective, Broadcom is forecasting that its AI chip business alone, in a single year, will generate more revenue than many of the largest semiconductor companies make in total.

Backing that confidence is a reported backlog in excess of $30 billion, much of it tied to multi-year custom-silicon commitments from hyperscale customers. Unlike spot GPU orders, custom-ASIC programs are designed years in advance and locked into product roadmaps, which gives Broadcom unusual visibility into future quarters. That visibility is exactly why management was willing to put a $56 billion full-year AI number on the record.

AI Semiconductor Revenue Trajectory

The following table tracks Broadcom’s disclosed and guided AI semiconductor revenue, illustrating the acceleration management is forecasting through fiscal 2026.

Period	AI Semiconductor Revenue	YoY Growth	Status
Q2 FY2026 (ended May 3)	$10.8B	+143%	Reported
Q3 FY2026 (guidance)	~$16B	+200%+	Guided
Full-year FY2026	~$56B	~+180%	Projected
Total semiconductor (Q2)	$15.0B	+79%	Reported
Trailing-12-month semis	~$48B	~75% AI	Reported

Why the Stock Fell After a Blowout Quarter

Here is the part that confused casual observers and crystallized something important for everyone else. Despite the beat and the raise, Broadcom shares sold off more than 10% in after-hours trading following the report, and one recap put the cumulative drop at as much as 13% as the broader market digested the news over the following sessions. The selloff was not isolated – it coincided with a broad Big Tech and AI risk-off move in early June 2026 that pulled high-multiple names lower together.

The explanations cluster around three themes. First, valuation. A stock that has compounded on the back of triple-digit AI growth carries expectations so high that even a strong quarter can fail to clear the bar the market has set. Second, the “tempered outlook” framing: some investors read management’s commentary on the infrastructure software segment and the pace of future AI scaling as a signal that the parabolic growth phase, while still strong, is beginning to normalize. Third, the macro backdrop – a sector-wide pullback in AI and Big Tech equities in early June that hit Broadcom regardless of its own fundamentals.

One earnings-reaction analysis captured the mood bluntly: the quarter was “a strong fiscal Q2 2026 earnings report” with “record free cash flow,” yet the stock “sold off more than 10% after hours.” Another commentator summarized the disconnect in real time – “Very good. So, yeah, double beat, but they’re down 6% or so right now.” The takeaway across the board was consistent: the problem was not the quarter. The problem was what the market had already paid for.

Broadcom’s Custom Chip Customers: Google, Meta and Beyond

The foundation of Broadcom’s AI franchise is its roster of custom-silicon customers – hyperscalers that co-design their own accelerators rather than buying off-the-shelf GPUs. The two most clearly identified in commentary around the quarter are Google, for which Broadcom is a long-standing partner on TPU silicon and attached networking, and Meta, tied to its MTIA accelerator program. These relationships are the engine behind the doubling-and-redoubling of AI revenue that management keeps describing.

The strategic logic for the hyperscalers is straightforward. A custom ASIC, designed for a specific workload, can deliver better performance-per-watt and lower total cost of ownership than a general-purpose GPU when run at the scale of a company training frontier models across hundreds of thousands of chips. By owning the design, the hyperscaler also reduces dependence on a single merchant supplier and gains leverage over its own roadmap. Broadcom monetizes that shift by providing the design IP, the advanced packaging expertise, and the networking silicon that stitches the clusters together.

Management has repeatedly pointed to additional prospective custom-chip customers beyond its established base, framing the pipeline as the source of future doublings in AI revenue. While Google and Meta are the names supported by the current reporting, the broader industry expectation is that the universe of companies designing their own accelerators will keep expanding as AI compute budgets grow. That pipeline – confirmed plus prospective – is what underwrites the $56 billion full-year target and the $30 billion-plus backlog.

Broadcom vs Nvidia: Two Models for the AI Boom

It is tempting to frame Broadcom and Nvidia as direct rivals, but they attack the AI compute market from different angles, and that difference is the most important thing to understand about both. Nvidia dominates the merchant GPU business – selling a standardized, programmable accelerator (backed by the CUDA software ecosystem) to anyone who wants it. Broadcom, by contrast, is the dominant enabler of custom silicon, helping a small number of hyperscalers build chips tailored to their own workloads, plus the networking that connects them.

These models are complementary as much as competitive. A hyperscaler can buy Nvidia GPUs for flexible, general-purpose training and inference while simultaneously deploying its own Broadcom-built ASICs for high-volume, well-defined workloads. The two approaches coexist inside the same data centers. Where they collide is at the margin: every dollar a hyperscaler spends on a custom accelerator is a dollar it does not spend on a merchant GPU, which is why the rise of custom silicon is watched so closely as a potential check on GPU dominance.

Dimension	Broadcom	Nvidia
Primary AI product	Custom ASICs (XPUs) + networking	Merchant GPUs
Customer model	Few hyperscalers, co-designed	Broad market, standardized
Software moat	Design IP, networking stack	CUDA ecosystem
Key advantage	Performance-per-watt at scale, TCO	Flexibility, time-to-market
Networking exposure	Tomahawk/Jericho switching, optics	NVLink, acquired networking
Reported AI revenue (latest qtr)	$10.8B	Not disclosed in this report

The honest answer to “who wins” is that, for now, both are winning. The total addressable market for AI compute has expanded fast enough to support a thriving merchant GPU leader and a thriving custom-silicon enabler simultaneously. The risk for investors in either name is the same risk that surfaced in Broadcom’s June selloff: the assumption that the growth never decelerates.

The Networking Story Hiding Inside the AI Numbers

One element that often gets lost in the headline AI revenue figure is how much of Broadcom’s edge comes from networking rather than raw compute. Training a frontier model is increasingly a distributed-systems problem: thousands of accelerators must exchange enormous volumes of data with minimal latency, or expensive compute sits idle waiting on the network. Broadcom’s Ethernet switching silicon, optical interconnects, and PCIe/connectivity portfolio sell into exactly that bottleneck.

This matters for the durability of the story. Custom ASIC programs can shift between vendors over time, and individual accelerator designs come and go. But the networking layer is sticky – it is embedded in the cluster architecture and tends to scale with the number of accelerators deployed, regardless of whose chips those are. In other words, Broadcom captures value whether a hyperscaler builds its own accelerators or buys merchant GPUs, because either way the cluster needs to be connected. That dual exposure is part of why management can speak with such confidence about multi-year growth.

Software Segment: Steady, But Not the Story

Broadcom is not only a chip company. Its infrastructure software segment – anchored by the VMware acquisition and a portfolio of enterprise software franchises – generated approximately $7.2 billion in revenue during the quarter, up about 9% year over year. That business throws off high-margin, recurring revenue that helps fund the company’s dividend and balance-sheet strength.

But 9% growth in a quarter where AI silicon grew 143% inevitably looks pedestrian, and some of the post-earnings caution traced back to commentary on the software segment’s pace. The market’s message was clear: investors are valuing Broadcom as an AI growth story first and a diversified infrastructure company second. When the AI line carries the multiple, anything that complicates the pure-growth narrative – even a steady, profitable software business – can become a source of caution rather than reassurance.

Historical Context: How Broadcom Became an AI Powerhouse

Broadcom’s AI ascent did not happen overnight, and understanding the arc helps explain why the company can sustain such confident guidance. For years, Broadcom was best known as a diversified semiconductor and infrastructure-software conglomerate assembled through a series of large acquisitions, with a reputation for disciplined capital allocation and elite margins rather than for cutting-edge AI. Its custom-silicon and networking businesses were respected but rarely the headline.

The generative-AI build-out changed that. As hyperscalers raced to add compute, two of Broadcom’s quietly dominant businesses – custom accelerator design and data-center networking – landed squarely in the path of the largest capital-spending wave the technology industry has ever seen. Revenue that once grew at modest rates began compounding at triple-digit percentages. The June 2026 quarter, with $10.8 billion in AI silicon revenue and a $56 billion full-year target, represents the culmination of that transformation: a company once defined by diversification is now defined by a single, enormous secular trend.

That history also explains the bull case for durability. Broadcom did not bet the company on AI; AI grew into the company’s existing strengths. The same discipline, margin focus, and cash generation that defined the pre-AI Broadcom now sit underneath a hypergrowth franchise – which is why free cash flow can run at 46% of revenue even while the top line nearly doubles.

Market Impact: A Bellwether for the Entire AI Trade

Because Broadcom sits at the intersection of custom silicon and networking, its results function as a read on the health of the entire AI capital-expenditure cycle. When Broadcom guides to $16 billion in quarterly AI chip revenue, it is implicitly confirming that hyperscalers are still committing enormous sums to build-out. That makes the report a bellwether – and it is why the post-earnings selloff rippled across the AI hardware complex rather than staying contained to one ticker.

The episode is best understood as a sentiment event layered on top of a fundamentals event. The fundamentals said: AI spending is accelerating, with multi-year visibility and a backlog north of $30 billion. The sentiment said: after a long run, the market is increasingly nervous about valuations and is quick to sell strength. Both things were true at once. For investors trying to read the broader AI cycle, the signal worth extracting is that demand remained robust even as risk appetite cooled – a divergence that tends to resolve in favor of fundamentals over time, though rarely on a predictable schedule.

What the Analysts Are Saying

Analyst commentary in the wake of the report leaned constructive on fundamentals even as the stock fell. The bull case rests on three pillars: the visibility provided by custom-silicon backlogs, the dual revenue capture across both accelerators and networking, and the cash-generation profile that lets Broadcom fund growth, dividends, and debt reduction simultaneously. One price target circulating after the report implied roughly 33% upside from prevailing levels, reflecting a view that the selloff was a multiple reset rather than a deterioration in the business.

The bear case is equally coherent. AI revenue cannot grow at 143% forever; at some point the law of large numbers and the maturation of hyperscaler capex will slow the pace. Critics also flag concentration risk – a custom-silicon business built around a handful of giant customers is exposed if any one of them shifts strategy or pulls in its spending. And the software segment’s 9% growth, while solid, does not offer the kind of offset that would cushion a deceleration on the AI side. The synthesis most analysts landed on: Broadcom is an exceptional business at a price that leaves little room for disappointment.

5 Predictions for Broadcom and the AI Chip Market

Based on the trajectory in the Q2 report and the structure of the AI build-out, here are five forward-looking predictions for the back half of 2026 and beyond:

1. The $56 billion full-year AI target holds or is raised. With a $30 billion-plus backlog and $16 billion guided for Q3 alone, Broadcom has structured the year so that the full-year number is well-supported by committed orders rather than speculative demand.
2. Custom silicon keeps taking share at the margin. As hyperscaler compute budgets grow, more workloads will migrate to purpose-built ASICs, expanding Broadcom’s design pipeline beyond its established base – though merchant GPUs remain dominant for general-purpose work.
3. Networking becomes a bigger part of the narrative. Expect Broadcom and analysts to increasingly break out the connectivity story, because it captures value regardless of which accelerators win and offers more durable, scale-linked growth.
4. Volatility stays elevated. With expectations this high, future quarters – even strong ones – are likely to produce sharp share-price swings as the market continually re-rates how much growth is already in the price.
5. Margins remain the differentiator. If Broadcom can keep adjusted EBITDA margins near 68-69% while scaling toward $29 billion-plus quarters, it will continue to stand apart from peers chasing growth at lower profitability.

Frequently Asked Questions

When did Broadcom report Q2 2026 earnings?

Broadcom reported its fiscal second-quarter 2026 results after market close on Wednesday, June 3, 2026. The quarter covered the period that ended May 3, 2026. The company’s next earnings report is expected on September 3, 2026.

How much revenue did Broadcom report in Q2 2026?

Total revenue was $22.187 billion, up 48% year over year and above the company’s guidance. Of that, semiconductor revenue was a record $15.0 billion (up 79%) and AI semiconductor revenue was a record $10.8 billion (up 143%). Infrastructure software contributed about $7.2 billion.

Why did Broadcom stock fall after strong earnings?

Despite beating estimates and raising guidance, shares sold off more than 10% after hours, with one recap citing a drop of as much as 13%. The decline was attributed to valuation concerns, cautious interpretation of the software-segment and AI-scaling commentary, and a broad AI/Big Tech market pullback in early June 2026 – not to any weakness in the reported numbers.

Who are Broadcom’s custom AI chip customers?

The customers most clearly identified in reporting around the quarter are Google (TPU silicon and attached networking) and Meta (its MTIA accelerator program). Management has also pointed to additional prospective custom-chip customers as the source of future AI revenue growth.

What is Broadcom’s full-year 2026 AI revenue forecast?

CEO Hock Tan projected approximately $56 billion in AI semiconductor revenue for full-year fiscal 2026, an increase of roughly 180% over fiscal 2025. For fiscal Q3 2026 specifically, the company guided to about $16 billion in AI semiconductor revenue, up more than 200% year over year.

How does Broadcom’s AI business compare to Nvidia’s?

They pursue different models. Nvidia leads the merchant GPU market with standardized, programmable accelerators backed by the CUDA software ecosystem. Broadcom enables custom ASICs co-designed with a small number of hyperscalers, plus the data-center networking that connects them. The two often coexist inside the same data centers, competing only at the margin where custom silicon displaces merchant GPUs.

What was Broadcom’s free cash flow in Q2 2026?

Free cash flow was $10.262 billion, equal to 46% of revenue. Cash from operations was $10.493 billion, and adjusted EBITDA reached $15.244 billion, or 69% of revenue – among the strongest margin profiles of any company at this revenue scale.

Related Coverage

• Intel Terafab Rally: 50% April Surge, $52 Stock [2026]
• Nvidia Ising: 35B-Param Quantum AI Model, 2.5x QEC Speed [2026]
• DeepX IPO: $700M Korean Edge AI Chip Bet [2026]
• Anthropic $65B Series H: $965B Valuation, IPO Looms [2026]
• GPT-5.5 Launch: 82.7% Terminal-Bench, $5 API [2026]
• AI Chips 2026: The Complete Market Guide

Sources & Further Reading

• Broadcom Q2 FY2026 Financial Results (PRNewswire)
• Broadcom (AVGO) Earnings History – MarketBeat
• Broadcom (AVGO) Stock Overview – MarketBeat
• Broadcom Inc. Official Site
• Nvidia Corporation Official Site

This article is for informational purposes only and does not constitute financial or investment advice. All figures are drawn from Broadcom’s fiscal Q2 2026 results reported June 3, 2026, and associated coverage. Last updated June 9, 2026.