Nvidia (NVDA) delivered blockbuster results on 20 May, reporting record revenue of $81.6 billion and crushing consensus forecasts. The shares fell 1.8% anyway. That tells you everything you need to know about where sentiment sits with this stock right now.
The earnings beat was undeniably impressive, with data centre revenue surging 92% year on year to $75.2 billion, while management guided the next quarter to $91 billion, another 95% increase, and the board also authorised an additional $80 billion of share buybacks, which on the surface gives the impression of unstoppable momentum.
Yet beneath the headline numbers, a more nuanced picture is emerging, with Nvidia shares up just 15% year to date in 2026, slightly ahead of the Nasdaq, while Intel has surged 76% on renewed foundry confidence and AI-driven server demand, and Micron has climbed 37% over the same period, suggesting Wall Street’s attention is starting to spread beyond the AI chip leader for understandable reasons.
The valuation question won’t go away
Nvidia trades on a price-to-earnings ratio of 41.28x, down from its 5-year median of 61x but still elevated relative to semiconductor peers. The price-to-sales ratio sits at 25.46x, more than five times the industry median of 6.05x. Analysts debate whether this is justified by growth, but investors are increasingly sceptical.
The forward price-to-earnings ratio of 25.77x appears more reasonable only in isolation. In context, it reflects expectations that earnings growth will continue at extraordinary levels for years to come. That is a high bar, and one the market is only now questioning after two years of unqualified enthusiasm.
Competition is arriving faster than expected
Nvidia still commands roughly 81% of the discrete AI chip market, a level of dominance that is undeniable, yet it is no longer unchallenged, with Advanced Micro Devices emerging as a credible alternative for cost-conscious hyperscalers by offering competitive performance at lower prices, while Intel is also working to rebuild its position in AI accelerators after years of missteps, supported in part by substantial government backing through the CHIPS Act.
More significant still is the shift towards custom silicon, with major cloud providers such as Microsoft, Google, Amazon and Meta increasingly developing proprietary AI chips for inference workloads, reducing reliance on Nvidia’s latest flagship processors for every use case, a trend that is gathering pace as margins come under pressure and customers grow more confident in alternative designs.
Nvidia’s leadership is aware of these shifts, noting on its first-quarter earnings call that inference, the deployment phase of trained models, is becoming the next key battleground, where cost efficiency and power consumption matter most, precisely the areas where Nvidia’s pricing power is least dominant, making this less of a short-term pressure point and more of a structural shift over time.
Supply constraints may be overstated
Nvidia reported a backlog of around 3.6 million units for its Blackwell chips, which the company presented as evidence of strong demand, a fair interpretation, although demand that outstrips supply does not automatically translate into a durable advantage, since the ability to sell ultimately depends on the capacity to manufacture and deliver those chips at scale.
Advanced node production at TSMC and other foundries faces hard physics constraints. Scaling is neither quick nor cheap. This means Nvidia faces a ceiling on near-term revenue growth, regardless of how much customers clamour for chips. Supply constraints may limit the magnitude of potential shortfalls, but they also limit upside surprise potential.
The macro backdrop is shifting
Kevin Warsh was sworn in as Federal Reserve Chair on 22 May, replacing Jerome Powell, a move that introduces uncertainty into the interest rate path, with officials signalling that rates may need to rise further to reach the 2% inflation target, and higher rates tend to compress the present value of long-duration growth stocks such as Nvidia.
Technology valuations have also begun to normalise after strong gains in early 2026, with the sector cooling as investors become less willing to pay premium multiples for growth at any price, creating a timing risk for Nvidia even if the underlying business remains strong.
Regulatory Obstacles are real
Export controls targeting advanced AI chip sales to China have forced Nvidia to modify products and delay launches. The H20 chip faces restrictions, and successors in the Blackwell generation designed for the Chinese market have been delayed. These constraints limit addressable market size and force the company to navigate a fragmented global regulatory landscape.
This is unlikely to improve, with geopolitical tensions around AI leadership continuing to harden rather than ease, forcing Nvidia to adapt its product portfolio across different jurisdictions, adding cost and operational complexity that is often overlooked in more optimistic market narratives.
What the numbers really tell us
Nvidia’s gross margins of 75% remain striking and point to pricing power that is still intact for now, but the shift from training to inference workloads is expected to place gradual pressure on those margins over time, and while that risk is not immediate, it is already beginning to matter in how the market assesses the long-term earnings profile.
The company’s order pipeline is substantial through 2027, reflecting confidence from major hyperscalers. Yet pipelines can shift, demand can soften, and competitive offerings can emerge. Visibility is genuine, but not ironclad.
Where the risk-reward lies
Nvidia remains the safest bet for investors betting on continued AI infrastructure spending. Its ecosystem advantage is genuine, and its execution has been faultless. The company earns every penny of its market dominance.
However, the market’s enthusiasm is beginning to price in several years of exceptional growth that may take longer to realise. Execution risk is rising just as valuation provides less margin of safety. Competitive intensity is increasing, regulatory headwinds are mounting, and the macroeconomic backdrop is less supportive.
For new money, the risk-reward profile looks materially less compelling than it did six months ago. For existing holders, patience with a consolidation phase may be the wisest approach. The story remains intact, but the price is catching up to the promise.
