AI chip expectations diverge: Nvidia beats expectations and Broadcom’s guidance misses, triggering a 1.3 trillion market value revaluation

From May to June 2026, the global semiconductor industry experienced a highly dramatic valuation restructuring. NVIDIA and Broadcom—two of the most representative companies in the AI chip sector—released their earnings reports within less than three weeks of each other, yet received completely different market reactions.

NVIDIA's FY27 Q1 revenue reached $81.6 billion, up 85% year-over-year, with data center business growing 92% YoY; Q2 revenue guidance of $91 billion exceeded market expectations. Although the stock price fluctuated briefly after the earnings release, the average target price from 62 Wall Street analysts remained as high as $298.

Broadcom's Q2 revenue for fiscal year 2026 was $22.19 billion, up 48% YoY, with AI semiconductor revenue of $10.8 billion, surging 143% YoY. However, the company's Q3 AI semiconductor revenue guidance was $16 billion, below the analyst consensus of $17.2 billion. This "miss" directly triggered a 10.3% single-day plunge of the PHLX Semiconductor Index on June 5, with the sector's market value evaporating over $1.3 trillion.

Similarly driven by AI performance growth, why does the market recognize NVIDIA's "better-than-expected" results, while responding to Broadcom's "high growth but slightly below expectations" with sell-offs? The core variable behind this is not the absolute value but the difference in expectation management.

Two earnings reports, two fates

NVIDIA: Surpassing expectations becomes the new normal

NVIDIA's FY27 Q1 data nearly broke records across all dimensions. Revenue of $13k, up 85% YoY, with a 20% quarter-over-quarter increase. Data center revenue of $75.2 billion accounted for over 92% of total revenue, up 92% YoY and 21% QoQ.

More noteworthy is the structural change. Data center computing revenue of $60.4 billion increased 77%; data center networking revenue of $14.8 billion soared 199%, also hitting a record high. This indicates that AI demand has expanded from single GPU computing power to a complete AI infrastructure—network interconnects, rack systems, optical communications, power, and cooling all becoming parts of the AI factory.

NVIDIA's guidance for FY27 Q2 revenue is $91 billion (±2%), significantly higher than the market consensus of $86–87 billion before the earnings. The company specifically pointed out that this guidance does not include data center computing revenue from China. This detail is extremely critical—being able to give such a strong outlook excluding the Chinese market suggests that global demand for AI computing power in other regions is sufficient to support high growth independently.

Broadcom: Impressive numbers, higher expectations

Broadcom's earnings are outstanding by any objective standard. Q2 total revenue of $22.19 billion, up 48% YoY; semiconductor revenue of $15 billion, AI revenue of $10.8 billion, up 143% YoY, and up 28% QoQ. CEO F. Y. F. Chen stated during the earnings call that AI semiconductor demand is "Simply insatiable."

In Q2, Broadcom's AI semiconductor orders exceeded $30 billion, while shipments during the same period were only $10.8 billion. The company expects full-year AI semiconductor revenue for FY26 to reach $56 billion, about 180% growth over FY25; for FY27, AI semiconductor revenue is expected to surpass $100 billion.

However, the market's core focus was on the Q3 AI semiconductor revenue guidance—$16 billion, below the analyst consensus of $17.2 billion. This $1.2 billion gap triggered a chain reaction.

Diverging logic of expectation management

The difference in market feedback for NVIDIA and Broadcom essentially stems from their philosophies of expectation management.

NVIDIA's strategy is "guide expectations then exceed them." Before the earnings, market consensus for NVIDIA's revenue was around $86.5 billion. The company responded with actual performance of $81.6 billion and guidance of $91 billion, not only surpassing expectations in the current quarter but also demonstrating sustained growth momentum with a guidance far beyond market forecasts for the next quarter.

Broadcom's approach is "deliver high growth but below the latest market expectations." In the five trading days before the earnings, Broadcom's stock price had already risen due to high market expectations for AI business. When the Q3 AI guidance of $16 billion was announced—below the $17.2 billion expected—the 7% gap, in a sector where "beat expectations" is the norm, was amplified and interpreted as a signal of potential weakness.

A deeper difference lies in the business predictability of the two companies. NVIDIA's general-purpose GPUs serve a broad range of AI training and inference scenarios, with customers including nearly all cloud providers and enterprise users, making demand highly visible and dispersed. Broadcom's custom ASIC/XPU business depends heavily on a few large clients—Google, Meta, ByteDance are core drivers. High customer concentration means greater revenue volatility, and the market demands more precise guidance.

$1.3 trillion chain reaction

The sell-off triggered by Broadcom's earnings quickly spread across the entire semiconductor sector. On June 5, the PHLX Semiconductor Index (SOX) plunged 10.3%, marking the largest single-day drop since the COVID-19 pandemic panic in March 2020.

Broadcom's stock itself fell about 14%, erasing roughly $286 billion in market value. NVIDIA dropped about 7%, falling below the $5 trillion market cap threshold. Micron Technology declined about 13%, AMD about 10%, Intel about 11%. In two trading days, the SOX index declined a total of 12%.

The catalyst was Broadcom's guidance miss, but the deeper driver was more complex. The semiconductor sector had experienced explosive gains for months, with the SOX index rising over 300% since late 2022. Valuations were at historic highs—VanEck's semiconductor ETF traded at a 67.2% premium over its intrinsic value. Meanwhile, strong US employment data heightened concerns about the Federal Reserve maintaining high interest rates, adding pressure on overvalued tech stocks.

This was a classic "high expectations + high valuation + small disappointment" systemic correction. Funds flowed out of AI semiconductor stocks into banks, industrials, and other traditional sectors, with the Dow Jones Industrial Average reaching a record high on the same day.

Structural evolution of the AI chip landscape

The sell-off triggered by the Broadcom event also reflects deeper changes occurring in the AI chip competition landscape.

NVIDIA currently holds about 75–80% of the AI chip market share, down from 87% in 2024, but still dominates. Morgan Stanley analysts note that NVIDIA's market share has "not diminished at all" over the past two years, maintaining its industry leadership. The CUDA ecosystem remains a formidable moat.

Broadcom, through its custom XPU/ASIC chip path, focuses on serving top-tier clients' specific needs. Google TPU, Meta MTIA, and OpenAI's in-house chip projects are all major orders for Broadcom. This model's advantages include high customer stickiness and attractive profit margins; disadvantages include high customer concentration and revenue volatility.

In the general GPU space, AMD directly competes with NVIDIA, with data center revenue of $5.8 billion in Q1 2026, up 57% YoY. However, in overall market share, there remains a significant gap compared to NVIDIA.

Morgan Stanley predicts that by the end of 2026, inference scenarios could account for 50–60% of AI chip demand, with NVIDIA's inference market share potentially dropping from 70% in 2025 to 60–65%. The rise of inference will provide greater growth opportunities for Broadcom and other custom chip vendors, but NVIDIA's dominance in training remains difficult to challenge in the short term.

Gate stock trading: a new channel to capture AI sector investment opportunities

For investors seeking to participate in AI chip sector investments, the Gate platform offers a stock trading service that provides a low-threshold, efficient pathway.

Coverage and trading methods: Gate's real stock trading includes over 11,500 stocks and ETFs, covering core AI chip targets like NVIDIA (NVDA), Broadcom (AVGO), AMD, Micron, as well as various industry-themed ETFs. Users can buy and sell within the familiar crypto trading interface without switching platforms.

Core advantages:

• Extremely low threshold for fractional trading: starting from 0.01 shares, with just $1 to begin US stock investment.
• USDT direct settlement: users can directly use USDT to trade US stocks, eliminating the cumbersome process of "selling crypto → withdrawing fiat → cross-border remittance → broker deposit."
• Compliance assurance: all stock trades are executed by compliant brokers holding US Broker-Dealer licenses and clearing qualifications, with full SIPC protection.
• Unified account management: crypto and stocks are managed within the same account, enabling more efficient asset allocation.

How to trade: After logging into the Gate platform, select the "Stocks" trading pair, enter the stock code (e.g., NVDA, AVGO), and buy or sell the corresponding stock at real-time market prices using USDT. The process closely resembles crypto trading, greatly reducing the learning curve for traditional US stock investing.

Conclusion

The divergence in market fortunes for NVIDIA and Broadcom in May–June 2026 exemplifies a typical "expectation management sensitive period" in the AI chip industry. When the entire sector's valuation has already fully reflected the optimistic scenario of "unlimited AI demand growth," any signals below expectations trigger disproportionate reactions.

NVIDIA, with its super-expectation-beating actual performance and even more optimistic next-quarter guidance, proves the sustainability of its growth momentum. Broadcom, despite delivering record-breaking quarterly results, saw a systemic sell-off triggered by a tiny Q3 guidance gap. This is not a fundamental reversal but a market recalibration under extreme pricing.

From a longer-term perspective, the growth narrative of the AI chip industry is far from over. Broadcom CEO F. Y. F. Chen reaffirmed during the earnings call that AI semiconductor revenue will exceed $1 trillion in FY27; NVIDIA continues to dominate with its full-stack GPU, networking, and CPU capabilities. These two companies represent two different technological routes and business models—general-purpose GPUs vs. custom ASICs—that will coexist and compete over the coming years.

For investors, short-term volatility in the AI chip sector should not obscure the long-term structural growth logic. On the Gate platform, users can leverage low-threshold fractional trading and USDT direct settlement to flexibly allocate core assets in the AI chip sector and seize this super cycle’s investment opportunities.