Key Takeaways
- Nvidia's Q2 revenue exceeded estimates, but its Q3 forecast and market reaction reflect high expectations.
- Concerns about AI spending deceleration are growing, compounded by China market uncertainties.
- Hyperscalers' in-house chip development could challenge Nvidia's data center growth.
- Nvidia's high valuation faces scrutiny amid potential growth tapering and market bottlenecks.
Deep Dive
- Nvidia reported Q2 revenue of $46.7 billion, exceeding estimates of $46.23 billion, with adjusted earnings per share at $1.05.
- Q2 data center revenue reached $41.1 billion, slightly below the $41.29 billion estimate, while gaming revenue surpassed expectations.
- For Q3, Nvidia forecasts revenue between $54 billion and $55.2 billion, surpassing the estimated $53.46 billion.
- The company also approved an additional $60 billion share buyback program.
- Nvidia faces challenges in China, where demand for future products like Blackwell is present but sales approval is pending due to US export restrictions.
- The company sold $180 million of H20 inventory, a deprecated AI chip version, outside of China.
- Analysts note uncertainty surrounding China's impact on future revenue, as demand is not as strong as last year.
- Fears of an AI spending slowdown are growing, with Nvidia's data center growth decelerating to mid-single digits from previous double-digit sequential growth.
- Hyperscalers are increasing capital expenditures, with Meta planning a 30% increase, which typically benefits Nvidia's data center growth.
- Concerns exist that other hyperscalers are developing their own chips, potentially reducing demand for Nvidia's products.
- Analysts Mandeep Singh and Jay Goldberg note Nvidia's high valuation, with a 40% earnings multiple implying significant future growth.
- Any tapering of this growth could lead to a multiple contraction for Nvidia's stock.
- Electricity availability is identified as a significant bottleneck for AI data centers.
- Nvidia's manufacturing reliance on TSMC also presents a potential constraint.
- Early issues with the reliability of Nvidia's new Blackwell systems have been noted by Jay Goldberg.
- Sarah Frier states that while AI investment continues, advancements in large language models may be slowing.