AIToday
AI Stocks & MarketsAI Business & IndustryYahoo Finance AIPublished: Aug 15, 2026, 19:00 JST5 min read

SpaceX commits exclusively to Nvidia, Wall Street says stock is undervalued

SpaceX commits exclusively to Nvidia, Wall Street says stock is undervalued

Key takeaway

  • SpaceX has committed to building exclusively on Nvidia's Vera Rubin superchip and expects to scale its AI compute capacity to several gigawatts by next year, positioning itself as a sixth major hyperscaler.

  • The move validates Nvidia's technology leadership in AI infrastructure and supports Wall Street's view that the stock is undervalued: with earnings projected to grow 45% annually over three years, Nvidia's price-to-earnings-to-growth ratio of 0.75 is the lowest in five years, and the median analyst target price of $300 implies 33% upside from the current $225.

3 Key Points

  1. What happened

    SpaceX reported 247% revenue growth in its AI segment in Q2, driven by cloud services deals with Alphabet and Anthropic. CEO Elon Musk announced SpaceX will build exclusively on Nvidia's Vera Rubin superchip and plans to reach over 2 gigawatts of compute by year-end, scaling to several times higher by end of next year. Wall Street analysts project Nvidia's earnings will grow 45% annually over the next three years, giving it a price-to-earnings-to-growth ratio of 0.75—the lowest in five years.

  2. Why it matters

    SpaceX's decision to use only Nvidia systems signals the company's confidence in Nvidia's technology and removes uncertainty about competing chip architectures. The commitment comes as SpaceX emerges as a potential sixth major hyperscaler (alongside Alphabet, Amazon, Meta, Microsoft, and Oracle); Morgan Stanley estimates SpaceX will invest $110 billion in AI infrastructure in 2028, representing annual growth of about 100%. Nvidia's 90% market share in data center accelerators makes it the cornerstone of the infrastructure buildout that hyperscalers are funding—Wall Street expects the top five alone to spend about $800 billion on AI infrastructure in 2026.

  3. What to watch

    Among 65 analysts, Nvidia has a median target price of $300 per share, implying 33% upside from its current share price of $225. Wall Street has repeatedly underestimated hyperscaler spending; the consensus estimate for top-five capex in 2026 rose from $525 billion last year to almost $800 billion now, suggesting analysts may also be underestimating Nvidia's future earnings growth.

In Depth

Read the full story

SpaceX's second-quarter results delivered validation for Nvidia investors. The company reported 247% revenue growth in its AI segment, driven primarily by cloud services deals with Alphabet and Anthropic and supplemented by enterprise AI tools. More significantly, CEO Elon Musk committed SpaceX to an exclusive partnership with Nvidia, stating on the earnings call: "We've decided to build exclusively on Nvidia because we think the Vera Rubin architecture is the best architecture. We think it's the best AI computer."

Vera Rubin is Nvidia's next-generation superchip architecture, pairing Vera CPUs and Rubin GPUs with chip-to-chip interconnects called NVLink. The module delivers about 10 times more performance per watt compared to its predecessor Grace Blackwell, making it substantially more efficient. Musk outlined an aggressive expansion timeline, telling analysts: "We expect to end this year with over 2 gigawatts of compute. And probably our cumulative compute online by the end of next year will be several times higher. It may, let's say, be closer to 10 gigawatts of compute than 5 gigawatts of compute." He added, "We're building AI compute capacity at scale faster than anyone else."

SpaceX's capital intensity signals its emergence as a major new player in infrastructure. The company reported $13 billion in AI-related capex in 2025 but has already exceeded that figure in the first half of 2026. Morgan Stanley estimates SpaceX's AI infrastructure investments will reach $110 billion in 2028, representing annual growth of about 100%. If accurate, SpaceX would join the top five hyperscalers—Alphabet, Amazon, Meta, Microsoft, and Oracle—as a sixth major investor in AI data center buildout. Goldman Sachs forecasts the top five alone will invest about $800 billion in AI infrastructure in 2026, with total AI-related capex topping $1 trillion.

Wall Street's valuation of Nvidia reflects cautious optimism anchored in earnings growth and a historical pattern of underestimation. Analysts forecast Nvidia's earnings will grow at 45% annually over the next three years. That growth rate, paired with the stock's current valuation of 33 times earnings, yields a price-to-earnings-to-growth ratio of 0.75—a level considered undervalued and the lowest the stock has achieved in five years. Among 65 analysts covering the name, the median target price is $300 per share, implying 33% upside from the current $225. The bearish case is complicated by a track record of capex underestimation: last year's consensus estimate for top-five hyperscaler capex in 2026 was $525 billion; the current estimate is almost $800 billion. If analysts have systematically underestimated infrastructure spending, they may have similarly underestimated the earnings growth that flows through to Nvidia.

Context & Analysis

Nvidia's dominance in AI infrastructure is reinforced by SpaceX's exclusive commitment to its chips. The company controls nearly 90% of the data center accelerator market and has diversified into networking and CPUs, making it the foundational layer of the hyperscaler buildout. SpaceX's decision to standardize on Vera Rubin eliminates architectural fragmentation and signals confidence in Nvidia's technical roadmap—a vote of confidence that may influence other large enterprises evaluating chip suppliers.

The scale of planned investment underscores why Nvidia's valuation may not yet reflect the reality of AI infrastructure spending. The top five hyperscalers (Alphabet, Amazon, Meta, Microsoft, and Oracle) are now forecast to invest about $800 billion in AI infrastructure in 2026, far exceeding last year's $525 billion estimate. Total AI-related capex is projected to top $1 trillion. SpaceX, which reported $13 billion in AI-related capex in 2025 but has already surpassed that through mid-2026, represents incremental demand beyond the traditional five. If Wall Street's capex estimates have lagged reality by this margin, earnings growth projections for Nvidia—currently set at 45% annually over three years—may similarly underestimate the company's medium-term revenue trajectory.

FAQ

What is Vera Rubin, and how does it compare to Nvidia's current chips?
Vera Rubin is Nvidia's next-generation superchip, featuring Vera CPUs and Rubin GPUs paired with chip-to-chip interconnects called NVLink. Compared to its predecessor Grace Blackwell, the Vera Rubin module delivers about 10 times more performance per watt, making it far more efficient.
How much is SpaceX planning to invest in AI infrastructure?
SpaceX reported $13 billion in AI-related capex in 2025, but has already surpassed that figure through the first half of 2026. Morgan Stanley estimates SpaceX's investments in AI infrastructure will hit $110 billion in 2028, representing annual growth of about 100%.
Why do Wall Street analysts think Nvidia is undervalued?
Wall Street estimates Nvidia's earnings will increase at 45% annually over the next three years, giving it a price-to-earnings-to-growth ratio of 0.75—values below 1 are typically taken to mean a stock is undervalued. Additionally, analysts have consistently underestimated hyperscaler capex spending; the 2026 estimate for the top five hyperscalers rose from $525 billion to almost $800 billion, suggesting they may also be underestimating Nvidia's future earnings.
Yahoo Finance AIRead Original Article

Get the latest AI Stocks & Markets news every morning

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytime

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleWorld Labs turns one robot task into thousands of simulated variants for training

The AI news that matters, in one minute each morning.

Sign up free