Who is leading the AI growth stocks in 2026? Analyzing the growth logic of NVIDIA, Broadcom, and Marvell Technology

As of June 15, 2026, the U.S. stock market's AI-centric technology growth stocks continue to lead the sector with a structural outperformance. The Nasdaq Composite closed at 25,888.84 points, and the Philadelphia Semiconductor Index (SOX) rose 1.52% on the day, closing at 13,371.47 points. In pre-market trading on June 15, the Philadelphia Semiconductor Index continued to strengthen by approximately 1.52%, while the Nasdaq maintained a high-level tone, with AI and semiconductor sectors still being the main directions of capital inflow. Against the backdrop of a generally bullish market with structural differentiation, the logic within the semiconductor sector is shifting from "widespread gains" to "performance realization."

In this AI computing power demand-driven rally, NVIDIA, Broadcom, and Marvell Technology continue to lead the sector with differentiated technological paths and business models. As of June 15, 2026, NVIDIA's latest quarterly revenue reached $81.6 billion, up 85% year-over-year, with growth accelerating for the third consecutive quarter; Broadcom's AI semiconductor business is expected to reach $56 billion for the full year, with a nearly 180% YoY increase; Marvell Technology is about to be included in the S&P 500 index, with a year-to-date increase of over 200%, attracting high market attention. Is the fundamental logic supporting their stock price expansion still robust? What do the latest events and data reveal? How does the newly launched stock trading feature on the Gate platform provide a pathway for allocating AI growth stocks?

NVIDIA: Revenue Accelerates, What Is the Waiting for a Trillion-Dollar Company?

NVIDIA remains the most solid player in the AI computing infrastructure market share. As of pre-market data on June 15, NVDA's market price is around $205, with Morningstar data showing the previous day's closing price at $208. The current valuation corresponds to an adjusted forward P/E ratio of approximately 31.42 times, with a total market capitalization of about $4.97 trillion.

From an earnings perspective, NVIDIA's fundamentals are still accelerating rather than slowing down. The latest quarterly revenue reached $81.6 billion, up 85% YoY, with growth accelerating for the third consecutive quarter (previous quarter 73%, two quarters ago 62%). Data center revenue grew 92% YoY to $75.2 billion, accounting for over 90% of total revenue. The company's guidance for the second quarter has been further raised to $91 billion. In terms of profitability, gross margin remains high at nearly 75%, significantly above most tech companies. The company also completed approximately $20 billion in buybacks and dividends in a single quarter and announced a new authorization of $80 billion for stock repurchase.

However, after the impressive earnings report, the stock price only rose slightly by 0.16% on June 12, with no market chasing higher. The core reasons are: first, the company's outlook assumes zero contribution from China's data center market, which previously contributed over one-fifth of data center revenue; second, large cloud service providers are accelerating procurement of NVIDIA chips while also rapidly developing alternative solutions. Goldman Sachs' latest report indicates that Microsoft, Google, Amazon, and Meta are expected to have combined capital expenditures of $770 billion in 2026, and the long-term erosion of ROE by AI capital expenditure is becoming an increasingly macro variable of concern for investors.

On the product side, NVIDIA announced several key developments at Computex in early June. The RTX Spark "super chip" was released, integrating Blackwell GPU with N1X/N1 CPU for PC use; the Blackwell platform has entered mass shipment, with customer feedback indicating increased procurement difficulty and longer delivery cycles, showing demand tension persists. The next-generation Rubin architecture has been sampled to some top partners, but large-scale shipments are expected only in the second half of FY2027. From supply-demand structure, NVIDIA has pre-locked capacity for DRAM and HBM, maintaining a leading position in supply chain layout.

Institutionally, 60 analysts maintain "Buy" or "Strong Buy" ratings, and 2,991 institutions have increased their holdings of NVIDIA, viewing it as a foundational asset in the AI era. However, over the past six months, there have been 98 insider sales and zero insider buys, indicating a clear divergence between management's outlook and institutional capital allocation logic.

Key focus: whether NVIDIA's guidance for the next quarter can sustain over 80% YoY growth will be crucial in determining if its valuation can hold current levels. If the $91 billion revenue target for Q2 is met as scheduled, the current P/E ratio of around 35 times, associated with high growth, is not "expensive"; but if cloud vendors' capital expenditure slows in the mid-cycle, valuation compression pressure will emerge.

Comparison of Core Growth Logic of Three AI Chip Giants

| Comparison Dimension | NVIDIA (NVDA) | Broadcom (AVGO) | Marvell Technology (MRVL) | | --- | --- | --- | --- | | Core Business Growth in 2026 | Data center revenue up 21% QoQ | AI semiconductor revenue up 143% YoY | Data center revenue accounts for about 75% | | Main Products or Services | Blackwell GPU (about 71% shipment share in 2026) | Custom ASIC/XPU (six key customers) | High-speed interconnect chips and custom ASICs | | AI-Related Revenue Guidance for FY2026 | Total revenue target raised quarterly, Q2 guidance $91 billion | Full-year AI semiconductor revenue $56 billion (about 180% YoY) | FY2027 full-year revenue guidance about $11 billion | | Institutional Capital Concentration | 2,991 institutions increased holdings, analyst buy ratings concentrated | 60+ institutions coverage, AI business as strongest growth engine | Inclusion in S&P 500 triggers index fund mandatory allocation |

Broadcom: Valuation Digesting After Record Results, Is the ASIC Leader’s Defense Still Stable?

Broadcom, pursuing a customized AI acceleration chip route, delivered a record-breaking second fiscal quarter on June 3, 2026: total revenue of $49.7k, up 48% YoY; AI semiconductor revenue of $10.8 billion, up 143%, hitting a new high. The second quarter added over $30 billion in AI semiconductor orders, far exceeding the actual shipment of $10.8 billion. For the entire FY2026, Broadcom expects AI semiconductor revenue to reach $56 billion, up about 180% YoY, and reaffirmed the FY2027 AI semiconductor revenue guidance exceeding $100 billion.

However, after this strong earnings report, AVGO's stock price fell about 13% in after-hours trading, closing at $382.07 on June 12, down 0.91% for the day. As of June 15, pre-market prices hovered around $382, with a year-to-date increase of about 9.91%, but the past 30 days saw a decline of about 10.14%. UBS maintained a "Buy" rating with a target price of about $485 (roughly 27% upside), and CITIC Securities also maintained an "Accumulate" rating with a target of $525. However, valuation remains debated, with the latest earnings data implying a P/E ratio in the 47–50 times range, with significant differences under various valuation frameworks.

Broadcom’s main defensive strategy is its long-term custom chip collaborations with Google TPU and Meta MTIA. Switching costs for custom ASICs are very high, with development cycles of 2–3 years, so short-term orders are unlikely to change. Notable changes are emerging from two directions:

First, the competitive landscape is shifting. NVIDIA CEO Jensen Huang publicly named Marvell Technology as "the next trillion-dollar company" at Computex, directly targeting Broadcom’s core areas. The market is reassessing the competitive ranking between Broadcom and Marvell in the custom ASIC market. The FY2026 report will be a key validation point: if Broadcom’s AI semiconductor revenue grows more than 15% quarter-over-quarter and order visibility extends beyond four quarters, the threat of substitution has not yet impacted orders; if growth slows below 20% for two consecutive quarters, market expectations will start to factor in competitive erosion.

Second, the impact of customers’ self-research strategies needs ongoing monitoring. Macquarie predicts Broadcom’s share of Google TPU-related revenue will decline from about 95% in 2026 to 65% in 2028. Nonetheless, Broadcom still has the most diversified customer base, including Google, Meta, OpenAI, and six other core custom chip clients, with orders extending into multiple future fiscal years, keeping short-term concentration risk within manageable bounds.

Core focus: the quarter-over-quarter growth rate of AI semiconductor revenue in the next earnings report is the most direct indicator of whether its competitive moat remains intact. If maintained above 15%, the multi-point distribution logic for custom ASIC demand holds; if growth slows significantly, the changing competitive landscape will begin to influence pricing models.

Marvell Technology: High Valuation, But Multiple Catalysts Still Coalescing

Among the three AI chip companies, Marvell Technology’s recent stock price and event catalysts are the most intense. As of June 15, MRVL’s market quote was around $279.70, down 0.36% for the day. Its total market cap is approximately $234 billion, with a year-to-date increase of over 200%, leading the entire AI growth stock sector.

The key event chain driving Marvell’s current rally is quite dense. At Computex on June 2, NVIDIA CEO Jensen Huang publicly stated that Marvell will be the next trillion-dollar company, emphasizing its critical value in AI data center interconnect bottlenecks—high-speed network switches, optical DSP, and custom ASICs, which are precisely the fastest-growing components during GPU cluster expansion. On June 5, the company announced inclusion in the S&P 500 index, which will take effect before the market opens on June 22, triggering mandatory allocation by index funds and supporting short-term structural capital inflows. On June 11, the stock surged 11.13% to $280.71 with a trading volume of 57.25 million shares, about 5.68% higher than the previous day. On June 15, the new CFO Dan Durn officially took office, replacing Willem Meintjes who served three years, and the company reaffirmed its FY2027 Q2 guidance of about $2.4 billion in revenue.

In terms of performance, FY2026 data center revenue exceeded $6 billion, about 75% of total revenue. The FY2027 Q1 revenue guidance is $81.6B, with an annual revenue target of about $11 billion. On the analyst side, B. Riley raised MRVL’s target price from $240 to $345 on June 12, maintaining a "Buy" rating, driven by deep cooperation with NVIDIA in NVLink Fusion, silicon photonics, and other areas; Stifel raised the target to $321; Benchmark set a target of $275. As of June 15, the consensus analyst target price ranges from $240 to $345, with current market prices approaching the upper middle of this range, indicating a high valuation.

From a valuation safety margin perspective, the current TTM P/E ratio of about 96–100 times is undoubtedly high. Institutions generally expect EPS to double over the next two years, driven by two main factors: the custom ASIC business is expected to surpass $10 billion in revenue by FY2029; the optical business is growing over 70% annually, supporting long-term growth. But high valuation implies high expectations for future growth, and any marginal downward revision in revenue or profit margins could lead to significant valuation correction.

Core focus: the incremental capital inflow after S&P 500 inclusion, whether next quarter’s guidance can exceed $2.5 billion, and whether new large-scale customer ASIC collaborations are disclosed—these key signals will determine the pace of Marvell’s "trillion-dollar" narrative shifting from expectation to reality.

Comparison of Core Valuation Indicators and Key Events of the Three AI Growth Stocks

| Comparison Dimension | NVIDIA (NVDA) | Broadcom (AVGO) | Marvell Technology (MRVL) | | --- | --- | --- | --- | | Recent Key Events | FY2027 Q1 revenue $81.6B, +85% YoY | Q2 AI semiconductor revenue $10.8B, new high | Inclusion in S&P 500 on June 22 | | Market Focus | Blackwell mass shipment, Rubin architecture in 2027 | Customer self-research evolution, gross margin trend | NVIDIA CEO’s "trillion-dollar" public endorsement | | Recent Valuation Metrics | TTM P/E approx 35.12x | TTM P/E approx 47–50x | TTM P/E approx 96–100x | | Recent and Forward Trends | Revenue structure changes as variable | FY2027 AI revenue guidance over $100 billion | High valuation relies on sustained high growth |

Gate Stock Trading: One-Stop Allocation of AI Growth Stocks with USDT

For investors optimistic about the AI track, directly completing low-threshold allocations of U.S. stock AI growth stocks within the crypto trading scene is a core pain point. Traditional U.S. stock trading paths typically involve multiple steps: "sell crypto → withdraw fiat → cross-border remittance → open brokerage account and deposit," often taking days. Gate TradFi’s real stock trading launched in June 2026 aims to compress this process to seconds.

In terms of asset coverage, Gate TradFi now offers over 10,000 real stocks and ETFs, covering all major U.S. exchanges including NYSE and NASDAQ. NVDA, AVGO, and MRVL are all available, allowing investors to concentrate their AI growth stock allocations without platform switching.

The core advantages of Gate stock trading are fourfold:

• USDT direct settlement: USDT in the account can be used directly to buy U.S. stocks, eliminating the need for fiat conversion and cross-border remittance, greatly improving trading efficiency.
• Fractional share trading with extremely low thresholds: starting from just 0.01 shares, about $1, investors can begin U.S. stock investments, lowering the capital barrier from hundreds of dollars to just a few dollars.
• Compliance and asset security: all stock trades are executed by licensed U.S. broker-dealers with clearing qualifications, assets are held independently in the DTC system, which is also a member of the SIPC, providing securities asset protection.
• Fee advantages: eligible users can enjoy fees as low as 0.023%, with no overnight financing or holding costs for spot trades. Investors holding real stocks have full rights to corporate actions, including dividends, stock dividends, rights issues, splits, and bonus shares, with dividends automatically credited to the Gate account.

Trading process: users update the Gate app to the latest version, go to “TradFi → Stocks,” transfer USDT from spot or unified accounts into the stock account, then search for the target stock and buy.

Conclusion

In the 2026 AI computing power race, NVIDIA, Broadcom, and Marvell Technology represent three differentiated growth paths: NVIDIA builds a technological moat with Blackwell and Rubin products, maintaining dominance in high-performance GPU markets; Broadcom relies on deep ties with large cloud providers through custom ASICs, supported by extremely high order visibility; Marvell emerges in high-speed interconnects and custom computing, gaining market attention through supply chain barriers and catalysts like index inclusion. All three have substantive, sustainable growth logic.

However, investors must also face structural risk nodes. NVIDIA faces long-term pressure from large customers’ self-research and substitution, and geopolitical factors’ ongoing impact on China market revenue; Broadcom needs to monitor the erosion of market share due to Google’s self-research push and the margin dilution from rising custom chip proportions; Marvell’s high valuation has already priced in substantial future growth expectations, and any delays in earnings realization could lead to significant valuation adjustments. Additionally, chip industry cyclicality, macro interest rate environment changes affecting high-valuation stocks, and geopolitical disruptions to supply chains collectively form a risk matrix for AI growth stock investments.

In this window of risk and opportunity, Gate’s stock trading feature provides investors with a low-threshold channel to participate directly in the global mainstream stock markets from the crypto ecosystem.