On June 3, 2026, Broadcom, the leader in custom AI chips, released a highly anticipated Q2 earnings report: revenue hit $22.187 billion, up 48% year-over-year; AI semiconductor revenue reached $10.8 billion, soaring 143% and setting a new record; adjusted EPS came in at $2.44, beating expectations. All three core metrics—overall revenue, EPS, and AI revenue—outperformed consensus estimates.
However, after the US market closed on June 4, AVGO shares plunged about 13%, dropping to around $418 in after-hours trading. This triggered a sharp sell-off in the Philadelphia Semiconductor Index, which fell over 10%, wiping out more than $1 trillion in market cap from the chip sector in a single day. The market doesn’t reward "good performance"—it rewards "performance that beats high expectations." In the seven trading days leading up to Broadcom’s earnings, its stock surged over 15%, adding roughly $300 billion in market cap. Investors had priced in "perfect expectations." When management guided Q3 AI chip sales to $16 billion—a year-over-year increase of over 200% but still below Wall Street’s consensus of $17.2 billion—and refused to raise the FY2027 AI revenue target ("over $100 billion" remains unchanged), it triggered the most sensitive sell-off switch for high-valuation stocks. Trading volume soared to 79.9 million shares, about 2.14 times the three-month daily average.
Q2 Earnings Review: Outperforming the Past, Underwhelming the Future
According to Broadcom’s official earnings release and management’s remarks during the June 3, 2026 earnings call, Q2 results were as follows:
- Total revenue: $22.187 billion, up 48% year-over-year, slightly above the consensus estimate of $22.13 billion
- Adjusted EPS (non-GAAP): $2.44, up 54% year-over-year, beating the expected $2.40
- AI semiconductor revenue: $10.8 billion, up 143% year-over-year; custom AI accelerators (XPU/ASIC) contributed $5.6 billion, AI networking chips accounted for $2.8 billion
- New AI semiconductor bookings for the quarter exceeded $30 billion, far above the $10.8 billion in actual shipments
- Adjusted EBITDA margin: 69%, beating the company’s previous guidance of 68%
- Infrastructure software revenue: $7.178 billion, up 9% year-over-year, slightly below the market estimate of $7.32 billion
For Q3 guidance, Broadcom expects total revenue of about $29.4 billion, up 84% year-over-year, beating the consensus estimate of $28.6 billion. Non-GAAP operating margin is expected to remain at 67%. However, the most closely watched AI chip guidance is $16 billion, below Bloomberg’s market average estimate of $17.2 billion—a gap of about $1.2 billion or 7%. Management further confirmed on the call that full-year 2026 AI semiconductor revenue is expected to reach $56 billion, up roughly 180% year-over-year, but still short of the market’s $57.7 billion forecast.
The long-term target of over $100 billion in AI semiconductor revenue for FY2027 remains unchanged. For investors who had priced in an "early upward revision," the lack of proactive guidance adjustment triggered a sell-off. When asked about margin pressure, CFO Kirsten Spears explained that as the share of custom AI chip shipments (like TPUs) rises, overall gross margin will narrow from 77% in Q2 to about 74%. This is fundamentally a shift in business mix, not a decline in efficiency.
Despite short-term stock volatility, the customer partnership details disclosed in Broadcom’s earnings are worth reviewing. Google signed a multi-generation TPU and AI networking long-term development and supply agreement in April. The Anthropic-related TPU access deal involves over 1 GW of compute deployment in 2026 and 5 GW annually starting in 2027. OpenAI’s end-of-2026 production agreement includes a commitment to deploy 1.3 GW in 2027, part of a total 10 GW partnership. For Meta, Broadcom expects to deploy 3 GW of MTIA XPU compute by 2028, with the first 1 GW order delivered in the second half of 2027. Long-term partnerships with six core custom chip customers remain intact, with orders extending into multiple future fiscal years.
Three Layers Behind the Sell-Off: Expectation Gap, Peer Comparison, and Macro Factors
On the surface, the Q3 AI revenue guidance missed consensus. But there are three deeper reasons.
First, front-loaded valuation excess. AVGO shares surged over 15% in the seven days before earnings, adding $300 billion in market cap and pricing in Q3 AI outperformance ahead of time. When actual results fell within Broadcom’s typically conservative guidance range—even though growth remained strong—it couldn’t support the elevated valuation. CFRA Research semiconductor analyst Angelo Zino’s core view: the after-hours drop mainly reflects sky-high pre-earnings expectations, not actual business deterioration.
Second, peer benchmark effect. Marvell Technology directly raised its AI ASIC revenue guidance in its May 27, 2026 earnings report. NVIDIA CEO Jensen Huang publicly endorsed Marvell at COMPUTEX 2026, saying it has trillion-dollar market cap potential. Marvell’s stock jumped over 30% in a single day after earnings. The market naturally compares: a competitor raises guidance, while the leader Broadcom opts not to.
Third, macro resonance. During Broadcom’s earnings release, US May CPI rose to 4.2% year-over-year—the highest since April 2023—fueling expectations for the Fed to maintain a hawkish stance. The S&P 500, Dow Jones, and Nasdaq all came under pressure. NVIDIA, Micron, and other semiconductor peers also fell, indicating this was a broad correction among high-valuation tech stocks, not a Broadcom-specific event.
The ripple effect extended to other semiconductor firms. Intel and AMD saw losses deepen over two consecutive sessions. Broadcom’s trading volume surged to 79.9 million shares as investors reassessed lofty AI growth assumptions. Macquarie downgraded Broadcom from "outperform" to "neutral," citing Google’s increased focus on developing its own AI chips, which could erode Broadcom’s ASIC market share over time. This concern triggered a market repricing of Broadcom’s customer concentration risk. Marvell’s progress in custom AI chips also intensified the competitive tension in the dual-leader landscape.
Understanding AVGO’s Post-Earnings Price Action
The stock has fallen steadily from its pre-earnings record high of $495, retreating about 24% as of June 11. On June 10, it dropped another 5.12% to close at $372.10, with trading volume at 38.19 million shares—up 3.15% from the previous session—showing continued selling pressure. On June 12, AVGO rebounded 3.62% to $385.57 intraday, reflecting some short-term relief but not a return to pre-earnings levels.
If Q3 delivers over 200% year-over-year AI semiconductor growth, and management reveals more customer contract details to enhance visibility, the market may reprice Broadcom’s long-term compound growth prospects. If competitive pressure shows up in future earnings as market share shifts or slower revenue growth—and customer concentration risk rises—valuation will face ongoing compression. Investors are tracking the next quarterly report in early September 2026 for actual AI semiconductor growth, customer expansion announcements, and updates on Google’s in-house chip project as key indicators.
From Single Event to Industry Structure: Revaluing the ASIC Track
Stepping back from individual earnings volatility, Broadcom’s "post-beat sell-off" signals a broader industry shift: the AI chip sector is moving from a GPU-centric narrative to a dual logic of GPU plus ASIC. The core driver is changing AI workload structure. According to Deloitte, inference workloads accounted for about one-third of total AI workloads in 2023, rising to two-thirds by 2026. Inference scenarios demand much higher compute efficiency per dollar and per watt than training, which is precisely the design focus of custom ASICs. For example, Google’s TPU delivers 3–5 times the per-watt performance of general-purpose GPUs in specific matrix operations.
Looking at scale, TrendForce expects AI ASIC chip shipments to grow 44.6% in 2026, compared to 16.1% for GPUs. Goldman Sachs forecasts ASICs will capture 40% of the AI chip market in 2026, surpassing 45% in 2027—nearly matching GPUs. Counterpoint Research projects AI ASIC shipments in 2027 will be triple those of 2024, and could overtake GPUs in 2028 with more than 15 million units shipped. In terms of market size, Goldman and others estimate the AI ASIC market will reach about $30 billion in 2026, rising to $70–80 billion in 2027—nearly half the industry. The paradigm shift from general GPUs to custom ASICs is becoming a new main theme for AI hardware investment.
In terms of competitive landscape, the ASIC market is led by Broadcom and Marvell. Broadcom holds about 70% of the custom AI chip market, with a client base including Google, Meta, OpenAI, Anthropic, and other hyperscale customers. Marvell focuses on custom projects like Amazon’s Trainium and Microsoft’s Maia, and has secured a key position in the NVLink-compatible ecosystem through hundreds of millions in strategic investment from NVIDIA. MediaTek has entered the smartphone SoC space, forecasting its AI ASIC business will contribute about $2 billion in Q4 2026 and expand to several billion in 2027, aiming for a 10–15% market share. Qualcomm is accelerating its data center custom chip product line through the AlphaWave acquisition. The ASIC supply landscape is evolving from a dual-leader model to a multi-player competition.
Gate TradFi: A New Cross-Asset Investment Channel from Crypto to Stocks
Against the backdrop of rapid change in the AI chip sector, investors need to follow the moves of core players like Broadcom, Marvell, and NVIDIA, while seeking more efficient, lower-barrier cross-asset allocation tools.
On June 1, 2026, Gate officially launched real stock trading services, becoming one of the first crypto platforms to directly offer access to the US stock market. As of June 2026, Gate TradFi has listed over 10,000 real stocks and ETFs, covering the five major exchanges: NYSE, Nasdaq, NYSE Arca, NYSE American, and BATS. The platform offers a wide range of assets from tech giants to thematic ETFs. Users don’t need to exchange currency, make cross-border transfers, or open separate brokerage accounts—simply use USDT in their Gate account to buy with one click. Gate’s US stock spot trading fees are as low as 0.023%, with VIP status starting at just $2,000. Investors holding real stocks enjoy all corporate actions, including cash dividends, stock dividends, rights issues, splits, and bonus shares, with dividend income automatically credited to their Gate account.
The recent launch of IPO Access further opens the door for users to participate in the early growth of high-potential companies. The first project is the highly anticipated SpaceX. Users can submit subscription requests before the company officially lists, pay with USDT, and receive shares directly in their Gate stock account after the IPO. This feature completes a full-chain investment product suite covering Pre-IPO, IPO subscription, and stock trading.
For crypto-native users who prefer round-the-clock trading, Gate offers two tokenized stock products—xStocks and Ondo Stocks—each pegged 1:1 to real stock prices, enabling 24/7 trading without traditional market hours restrictions. Both products can be flexibly chosen based on allocation needs, allowing unified account management across traditional finance and digital assets.
Conclusion
The sharp reaction to Broadcom’s Q3 AI guidance of $16 billion—below market expectations—should not be seen as a sign that AI infrastructure investment has peaked. A more accurate interpretation is that capital markets are shifting from "chasing high growth across the board" to "verifying growth quality one by one," with marginal expectations becoming the key pricing variable. The custom AI chip sector’s rapid growth trend remains intact. As inference workloads expand and hyperscale cloud customers pursue optimal TCO, ASIC’s structural market share is rising at a steady pace.
Gate’s newly launched real stock trading service integrates US equities into a unified account system, complemented by IPO Access and other innovative features, providing investors with a comprehensive channel for core AI-themed assets and cross-asset allocation. As Broadcom undergoes this expectation reset, the market will focus more on actual AI business growth in upcoming earnings, new customer developments, and competitive positioning. How these factors evolve will determine the next valuation anchor for the ASIC dual leaders. Investors should closely monitor Q3 actual growth, customer expansion, and competitive dynamics, seeking allocation strategies that match their risk preferences amid high volatility.
