Broadcom (NASDAQ: AVGO) is a global semiconductor and infrastructure software company that designs custom AI accelerator chips and high-speed networking components for some of the world’s largest technology firms. On June 4, 2026, shares of Broadcom plunged more than 15% in a single session after the company’s third-quarter AI chip revenue guidance fell short of Wall Street’s expectations. The selloff erased roughly $350 billion in market capitalization and sent shockwaves through the broader chip sector, dragging down names like AMD, Intel, Micron, and Marvell along with it.
In this article, we’ll discuss what happened during Broadcom’s fiscal Q2 2026 earnings report, why the market reacted so harshly despite the company technically beating several consensus estimates, what their “chips only” strategic pivot means, and how this moment fits into the larger narrative around AI infrastructure spending. We’ll also explore what this selloff signals for investors, the semiconductor industry, and the future of custom AI silicon.
TL;DR Snapshot
Broadcom reported record Q2 2026 revenue of $22.2 billion and earnings that topped estimates, but its third-quarter AI chip revenue guidance missed analyst expectations. Despite 143% year-over-year AI revenue growth, the company’s decision not to raise its full-year AI semiconductor target triggered a dramatic selloff that rippled across the global chip sector.
Key takeaways include…
- Broadcom’s Q3 AI chip revenue guidance of $16 billion missed the analyst consensus of $17.2 billion, and the company held firm on its full-year $100 billion AI semiconductor forecast rather than raising it, disappointing investors who expected an upward revision.
- The 15% single-day stock drop erased roughly $350 billion in market value and dragged other semiconductor names lower, with AMD falling 4%, Micron and Marvell dropping around 7%, and Intel slipping 3%.
- CEO Hock Tan announced a strategic shift to “chips only,” stepping away from Broadcom’s previously announced plan to sell complete integrated AI systems, signaling a narrower but potentially more focused business model going forward.
Who should read this: Investors, financial analysts, tech industry professionals, semiconductor enthusiasts, and anyone following the AI infrastructure boom.
The Numbers Were Good, But Not Good Enough
Here’s the paradox at the center of Broadcom’s selloff: by most conventional measures, the company delivered a strong quarter. According to CNBC’s earnings report, revenue climbed 48% year-over-year to a record $22.19 billion, and adjusted earnings per share came in at $2.44, beating the consensus estimate of $2.40. Net income surged 88% to $9.31 billion. AI semiconductor revenue specifically hit $10.8 billion, growing 143% compared to the same quarter a year ago. Those are objectively impressive numbers for a company operating at this scale.
But the AI chip market doesn’t necessarily reward “impressive,” especially not when there are concerns with a company’s forward guidance. Broadcom had projected $16 billion in AI chip revenue for the fiscal third quarter, a figure that would represent more than 200% year-over-year growth. Under normal circumstances, tripling your AI revenue would be cause for celebration. But analysts had been expecting $17.2 billion according to Yahoo Finance, and that $1.2 billion gap between guidance and expectations was enough to trigger a massive selloff. Making matters worse, CEO Hock Tan chose not to raise the company’s full-year AI semiconductor revenue target, reiterating the existing guidance of “in excess of $100 billion” rather than bumping it higher. As CFRA Research senior VP Angelo Zino told 24/7 Wall Street, “The bar was really high going into the print here, and I think part of the response you’re seeing here from the shares kind of points to that.”
Dan Coatsworth, head of markets at AJ Bell, offered a similar assessment. As reported by Yahoo Finance, Coatsworth noted Broadcom is finding that meeting and even slightly beating forecasts isn’t enough when the market holds it to such a high standard. With the stock up 88% over the prior year and roughly 55% in the quarter alone leading into earnings, there was simply no room for anything less than a blowout.
The “Chips Only” Pivot and What It Signals
Beyond the numbers, Broadcom’s earnings call introduced a strategic shift that caught some investors off guard. According to CNBC, CEO Hock Tan announced that Broadcom would offer “chips only” going forward, stepping back from its previously stated plan to provide complete integrated AI systems to its customers.
This is a meaningful change. Broadcom had been positioning itself as more than just a chip designer, moving toward a model where it would deliver full AI compute solutions, including the chips, the networking, and the surrounding infrastructure. Walking that back suggests a more focused strategy, but it also implies that the margin opportunity from selling complete systems won’t materialize as some investors had anticipated.
Tan also provided insight into customer behavior, noting on the call that bookings aren’t necessarily for immediate delivery. “Some they hope to have, but the reality they all accept is they need to align quite a few other things in place before they can deliver,” he said, according to CNBC. This suggests that while demand for AI chips remains strong, there’s a growing gap between the pace at which companies are ordering silicon and the pace at which they’re actually deploying it at scale.
Broadcom currently counts six core custom chip customers, including Google, Meta, Anthropic, and OpenAI, as reported by CNBC. The company recently expanded a custom chip partnership with Meta through 2029 and signed a compute agreement with Google and Anthropic covering approximately 3.5 gigawatts of capacity, per Yahoo Finance. These are long-term, high-value relationships, but they also mean Broadcom’s revenue concentration is significant. If even one or two of these customers slow their deployment timelines, the ripple effects on guidance could be substantial.
The Ripple Effect Across Semiconductors
According to GoTrade News, the Broadcom hit immediately dragged other chip stocks lower as well. Intel and AMD slipped between 1.5% and 4%. Qualcomm fell about 4%, while Micron and Marvell dropped around 7%, as reported by Yahoo Finance. S&P 500 futures dropped 0.5% and Nasdaq 100 futures weakened by 0.7% in early trading.
The broader market context made things even worse. According to Yahoo Finance’s coverage, the tech sector was the biggest laggard at the open, while Bitcoin also fell 4% to around $64,000 as investors adopted a risk-off posture. Geopolitical tensions related to rising instability in the Iran region added further pressure on sentiment.
Analyst Ryan Lee of Direxion captured the mood, stating that the market now demands perfection for the AI chip rally to continue, as reported by GoTrade News. This is the challenge for the entire semiconductor sector in mid-2026: the AI trade has been so powerful and so sustained that anything short of accelerating upside gets treated as a warning sign.
What This Means for the AI Infrastructure Boom
It’s worth stepping back to put Broadcom’s results in the context of the broader AI spending wave. The major hyperscalers, including Microsoft, Alphabet, Amazon, Meta, and Oracle, are on track to spend upward of $650 billion on AI-related infrastructure in 2026, according to Yahoo Finance. That’s a roughly 67% increase from the $381 billion these companies spent in 2025 and represents what some analysts have called the largest peacetime industrial mobilization in U.S. history, as described in a Futurum Group analysis.
This spending is real, it’s accelerating, and Broadcom sits squarely in the middle of it. The company designs custom AI accelerators and networking chips for exactly the kind of massive data center buildouts these hyperscalers are undertaking. The fact that Broadcom’s AI revenue is still growing at 143% year-over-year, and is projected to grow over 200% next quarter, underscores that the fundamental demand picture hasn’t changed.
What has changed is investor expectations. The AI chip trade has been one of the most profitable themes in the market since late 2022, when ChatGPT kicked off the generative AI boom. Broadcom’s stock had multiplied almost ninefold from the end of 2022 through mid-2026, as noted by CNBC. At those kinds of valuations, the market isn’t pricing in a good business. It’s pricing in a continuously accelerating one. Any deceleration, even from 143% growth to “only” 200% projected growth, gets punished.
The supply side adds another layer of complexity. As Astute Group notes, Broadcom had previously flagged that constrained foundry capacity at TSMC would limit chip supply into 2026. Advanced packaging capacity, particularly TSMC’s CoWoS technology, remains a secondary bottleneck. Even with massive demand, there are physical limits to how quickly new chips can be manufactured and delivered.
For investors, the Broadcom selloff is a reminder that in a momentum-driven market, execution needs to exceed expectations, not just meet them. For the AI industry more broadly, the spending commitments from hyperscalers remain enormous and growing. The infrastructure buildout isn’t slowing down, even if individual earnings reports create turbulence along the way.
Frequently Asked Questions
Broadcom Inc. (NASDAQ: AVGO) is a global technology company headquartered in Palo Alto, California. It designs, develops, and supplies semiconductor devices and infrastructure software solutions. In the AI space, Broadcom is best known for designing custom AI accelerator chips (sometimes called ASICs or XPUs) and high-speed networking components for major technology companies.
Custom AI accelerators are specialized processors designed for a specific company’s AI workloads, as opposed to general-purpose GPUs. Broadcom works with clients like Google (which uses Broadcom-designed tensor processing units, or TPUs), Meta, and others to create silicon tailored to their particular training and inference needs.
A hyperscaler is a company that operates massive-scale cloud computing and data center infrastructure. The term typically refers to the largest technology companies, including Amazon (AWS), Microsoft (Azure), Alphabet (Google Cloud), Meta, and Oracle. These companies are collectively spending hundreds of billions of dollars per year on AI-related infrastructure.
Broadcom had previously indicated it would move toward selling complete integrated AI systems, combining its custom chips with networking and other infrastructure into a turnkey solution. The “chips only” pivot means Broadcom is stepping back from that approach and will instead focus on delivering just the semiconductor components, leaving system integration to its customers.
TSMC (Taiwan Semiconductor Manufacturing Company) is the world’s largest contract chip manufacturer. Most advanced AI chips, including those designed by Broadcom, Nvidia, and AMD, are fabricated at TSMC’s facilities. When TSMC’s capacity is constrained, it creates a bottleneck that can limit how many chips companies like Broadcom can deliver, regardless of how much demand exists.
Broadcom has set a long-term revenue target of exceeding $100 billion from AI semiconductors. This figure covers the company’s outlook across fiscal years 2026 and into 2027. Investors were disappointed that Broadcom didn’t raise this target despite strong Q2 results, interpreting the decision as a sign that management isn’t confident in accelerating growth beyond current projections.
Other Enterprise AI Articles You May Be Interested In
Publishers Can Now Opt Out of Google AI Overviews: Inside the UK’s Historic Ruling
Claude Opus 4.8: Everything You Need to Know About Anthropic’s New AI Model
Mailchimp Launches Analytics AI: What Ecommerce Brands Need to Know
CNN Sues Perplexity AI Over Copyright Theft: What You Need to Know
Disarm AI: Inside Pope Leo’s Encyclical and Its Challenge to the Trump Administration’s AI Strategy
