Nvidia Stock Won't Be Overvalued by 2028: My Case for Buying NVDA Today

Nvidia is projected to nearly double revenue in fiscal Q2 (vs. an expected 97% revenue growth acceleration after 85% growth in Q1) as it launches its Rubin platform. The article highlights a low PEG ratio of 0.62 (as of Aug 2026) suggesting the stock is undervalued despite a trailing P/E of 37.5, with profits potentially doubling from 2026 to 2028. It also cites hyperscaler AI capex of roughly $700B this year (rising next year) and SpaceX targeting 10GW of compute by end-2027—supporting continued demand for Nvidia data-center GPUs.

Analysis

The cleanest market read is not just "NVDA up," but that AI capex is becoming a larger tax on the rest of the ecosystem. If spend keeps compounding at the pace implied here, the marginal winners are not only NVDA but also the supply chain with pricing leverage and scarce capacity — TSM, ASML, and advanced-packaging/power infrastructure names — while the first-order losers are the hyperscalers if AI monetization lags the capex curve. That creates a delayed margin squeeze risk for AMZN, MSFT, META, and GOOGL: revenue can reaccelerate, yet free-cash-flow multiples can still compress if investors start questioning payback periods.

The near-term catalyst is not the long-dated narrative; it is the next 1-2 earnings prints and, more importantly, guidance on data-center backlog, gross margin, and supply availability. The biggest falsifier is any sign that customers are digesting prior orders rather than accelerating them, because this type of trade is usually driven by sequential order revision, not the headline growth rate. Conversely, if Rubin ramps cleanly and margins hold, the market may keep paying for a longer duration growth stream than a typical semiconductor cycle deserves.

The contrarian miss is that the market often confuses large announced capex with durable earnings power. If the buildout is real, some of the alpha migrates away from NVDA into the less obvious picks-and-shovels names; if it is not, NVDA’s premium multiple is the first to compress because expectations are highest there. SpaceX and startup IPO talk is optionality, not cash flow — the tradeable issue is whether hyperscaler demand stays elastic after another quarter of aggressive spending.

Key Decisions for Investors

• Buy NVDA on 3-5% pullbacks or post-earnings weakness; hold 3-6 months. Risk/reward is favorable if data-center growth stays above ~50% y/y and gross margin does not sequentially deteriorate.
• Pair trade: long NVDA / short AMD on AI-capex momentum. Thesis: NVDA keeps pricing power and ecosystem lock-in while AMD is more exposed if buyers become more selective; exit if AMD design-win commentary accelerates or NVDA guidance softens.
• Use NVDA call spreads rather than outright stock if implied volatility is elevated. Structure for a 3-6 month horizon; the trade works if the market continues to re-rate AI revenue durability, but it should be abandoned on any guidance cut or margin miss.
• Watch AMZN, MSFT, META, and GOOGL capex commentary as a reverse indicator. If combined AI capex growth slows materially versus current run rate, reduce NVDA exposure because the stock is pricing sustained buyer urgency, not just current demand.
• If looking for lower-beta AI exposure, consider TSM or ASML over NVDA on weakness. They benefit from the same buildout with less single-name execution risk, though upside is less convex than NVDA.