Nvidia Just Returned a Record $26 Billion to Shareholders After Increasing Its Dividend by 2,400%. I Predict Another Massive Dividend Increase.

Nvidia’s Q2 FY2027 results showed revenue and operating income more than doubling YoY while holding a 75% gross margin despite a 55% rise in operating expenses. The company returned a record $25.78B to shareholders in the latest quarter ($20B buybacks + $6.05B dividends) and is targeting 60% of YTD free cash flow returns vs its 50%+ plan, with a 2,400% dividend jump to $0.25/share. Looking ahead, Vera Rubin shipments began in August and management expects it to drive ~20% of data center revenue in the next quarter, alongside fiscal 2028 revenue guidance of +70% YoY—supporting further dividend increases and accelerated buybacks.

Analysis

The real equity signal is not the dividend hike; it is that NVDA is converting AI scarcity into a capital-return flywheel while still re-rating the product cycle. That combination usually supports the multiple near term because buybacks become an earnings-per-share backstop, but it also shifts where excess value accrues: into NVDA’s own stock and into the infrastructure vendors that sit inside the rack-scale bill of materials. The clean second-order winners are HBM / memory and networking names such as MU and ANET; the cleanest losers are accelerator share challengers like AMD and, longer term, any custom-ASIC effort that cannot match NVDA’s platform integration.

The catalyst path is front-loaded over the next 1-3 months: investors will focus on whether Rubin ramps without margin dilution and whether management keeps nudging capital returns higher as FCF expands. The medium-term risk is that a capital-return story can be read as maturity before it is actually saturation; if hyperscaler capex growth normalizes or memory inflation persists, the market could decide the cash is real but the growth runway is less durable. Falsifiers are simple: Rubin shipments missing the implied contribution, gross margin cracking materially, or data-center growth decelerating faster than FCF per share.

Consensus may be underestimating how much of the upside is already embedded in the platform migration, which makes the best risk/reward likely relative rather than absolute. NVDA still looks like the highest-quality expression of AI capex, but after a strong run the cleaner trade is to own NVDA against something with weaker pricing power and less visible cash conversion. If Rubin proves to be a supply-chain multiplier rather than a demand pull-forward, the downstream beneficiaries can outperform the headline stock on a 3-6 month lag.

Key Decisions for Investors

• Maintain a core long NVDA position, but add only on 3-5% pullbacks; thesis works best over 3-12 months if Rubin drives sequential guide raises and free cash flow keeps scaling. Falsify on any meaningful gross-margin compression or a delayed Rubin ramp.
• Pair trade: long NVDA / short AMD for 1-3 months to express platform-share consolidation and relative margin durability. Best if you expect the market to reward full-stack control over standalone accelerator exposure; cover if AMD re-accelerates revenue growth or wins incremental cloud sockets.
• Initiate or add to MU on weakness as the highest-beta second-order beneficiary of AI rack content over 6-18 months. Risk/reward improves if HBM pricing stays firm and if NVDA’s rack-scale adoption broadens; thesis breaks on a DRAM/HBM price roll-over.
• Buy ANET as a lagged beneficiary of rack-scale networking content for a 3-6 month trade. This is the cleaner way to express NVDA ecosystem expansion without paying the headline multiple; invalidated if enterprise AI capex pauses or Ethernet switching competition compresses margins.
• No fresh short on NVDA unless data-center growth inflects down: the capital-return backstop makes outright bearish positioning unattractive near-term. If you need a hedge, use SOXX or a smaller AI semis basket instead of single-name shorts.