Nvidia forecasts revenue above estimates, announces $80-billion share buyback program

Nvidia forecast Q2 revenue of $91 billion, plus or minus 2%, above the $86.84 billion Wall Street estimate, and authorized an $80 billion share repurchase program. The company also raised its quarterly dividend from $0.01 to $0.25 per share, reinforcing capital returns, even as shares fell more than 2% in extended trading. The outlook highlights continued AI demand, though rising custom-chip competition from Big Tech and rivals such as Intel and AMD remains a structural risk.

Analysis

The market is treating this like a classic “good news, bad reaction” setup, but the more important signal is that hyperscaler capex is still intact while the mix is shifting toward self-supply. That is a subtle but meaningful change: if the largest buyers increasingly internalize inference workloads with custom silicon, Nvidia’s revenue pool can keep growing even as unit economics and share of wallet get pressured over the next 6-18 months. The buyback is also a tell that management sees a strong near-term cash conversion profile, which should put a floor under the stock, but it does not neutralize the strategic risk from customer vertical integration.

The immediate winners are the cloud platforms, not the semiconductor vendors. Alphabet, Amazon, and Microsoft gain leverage because custom chips improve their cost per token and reduce dependence on an external supplier that has been able to tax the entire AI stack; over time that should show up first in inference margin expansion, then in pricing power for AI services. The losers are the inference-exposed chip challengers: Intel and AMD can still win selective sockets, but the addressable market they are targeting is likely to become more price-competitive exactly as hyperscalers standardize on in-house silicon, which compresses the economic value of “good enough” accelerators.

The contrarian takeaway is that the current move may be overfocused on the headline beat and underfocused on durability of earnings quality. In AI semis, the market tends to pay up for visible demand, but the real risk is a multi-quarter transition from scarcity pricing to procurement discipline as inference scales and customers gain negotiating leverage. That means the next inflection is not shipment volume but gross margin trajectory and commentary on attach rates for custom silicon versus merchant GPUs. If those comments soften, the stock can derate quickly even if top-line growth remains exceptional.

Near term, the catalyst path is asymmetric: a few sessions of relief if buybacks/dividends dominate the tape, but over 1-2 quarters the key risk is that hyperscaler capex stays high while Nvidia’s share of that spend leaks to internal chips. That combination would be bearish for relative performance versus the big buyers, while also narrowing the long-only case for AMD and INTC unless they can demonstrate inference wins outside the hyperscaler ecosystem. The setup favors relative-value expressions over outright directional longs, because the structural AI spend thesis remains intact even as the beneficiary set broadens.

Key Decisions for Investors

• Long MSFT/GOOGL/AMZN vs short NVDA on a 3-6 month horizon: own the buyers of AI infrastructure as they capture inference economics; risk is NVDA reaccelerates margin expansion and the pair widens against us.
• Short INTC and AMD into strength for 1-3 months: the market is likely to overestimate near-term inference share gains; use a rally to establish positions with tight risk controls, as any announced design win can create sharp squeezes.
• Buy NVDA downside protection via 1-2 quarter puts rather than shorting outright: the stock has support from cash returns, but a guide-down in gross margin or commentary about custom-chip penetration could trigger a fast multiple reset.
• If long NVDA, pair it with a smaller long in a hyperscaler basket: this keeps exposure to AI capex intact while reducing single-name risk that merchant GPU share gets diluted over the next two quarters.