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

Nvidia guided Q2 revenue to $91 billion, plus or minus 2%, above the $86.84 billion consensus, and authorized an $80 billion share repurchase program. It also raised its quarterly dividend to 25 cents from 1 cent per share, reinforcing capital returns. The report is positive for AI spending trends, though shares were slightly down 0.2% after hours amid growing competition from custom chips and rival silicon.

Analysis

The market is still treating NVDA as a pure semiconductor name, but the bigger signal is that its economics are migrating toward platform behavior: buybacks and a higher dividend imply management sees free cash flow durability even if unit growth normalizes. That matters because the key debate is no longer whether AI capex keeps growing this year, but whether NVIDIA can defend share as workloads shift from training to inference, where pricing power is structurally lower and customers have more incentive to dual-source.

Second-order winners are the hyperscalers, not the chip vendors. If custom silicon adoption keeps improving, GOOGL/AMZN/MSFT can compress their inference cost curve and improve cloud gross margin, even if headline capex stays elevated; the market likely underappreciates how quickly this can offset slower external GPU growth. The losers are the inference challengers—AMD and INTC—because they need not just better chips but an ecosystem win, and NVIDIA’s move into CPUs/system-level integration is an attempt to attack the sockets where procurement decisions are made, not just the accelerator slot.

The contrarian risk is that investors overread the buyback as a confidence signal and underread it as a capital-allocation response to a narrowing reinvestment set. If hyperscalers prove they can internalize more of the stack over the next 12-24 months, NVIDIA’s growth can stay strong while multiple expansion stalls. The catalyst to watch is the next 2-3 quarters of cloud capex commentary: if inference spend shifts from external GPUs to proprietary silicon faster than expected, the current leadership rotation could reverse into a relative performance trade against NVDA and into the hyperscalers.

Key Decisions for Investors

• Go long GOOGL/MSFT vs short NVDA in a 3-6 month pair trade. Thesis: hyperscalers capture more economics as inference is internalized; target 10-15% relative outperformance if custom silicon rollout continues, with stop-loss on any renewed acceleration in external GPU demand.
• Buy AMZN on dips over the next 1-2 months as a beneficiary of lower inference COGS. Risk/reward is attractive because margin expansion can come from inside the model rather than top-line reacceleration; downside is only if capex intensifies without efficiency gains.
• Short AMD on strength for a 6-12 month view. It needs share gains in a market where the incumbent is expanding into CPUs and integrated systems; the asymmetry favors maintaining NVDA ecosystem dominance over a clean disintermediation story.
• Consider a tactical long NVDA / short SMH hedge only on pullbacks, not breakouts. The stock still has fundamental support from cash generation, but upside likely lags the quality of the guidance unless the market re-rates the inference opportunity.
• Use 6-12 month call spreads on GOOGL or MSFT instead of outright longs if you want convexity to AI cost savings. That captures the possibility that internal silicon becomes a margin catalyst without paying for a full multiple expansion.