Not Nvidia. Not Palantir. This AI Stock Could Have the Most Upside.

The article questions whether AI stocks are in a bubble due to “lofty premiums,” but argues valuations can stay elevated if growth executes. It highlights SpaceX (Space Exploration Technologies) at a reported $2T valuation and claims SpaceX’s IPO TAM is $28.5T (including $26.5T in AI), implying potentially large upside if SpaceX can manufacture GPUs and build space-based data centers. However, it flags major execution risk—Morningstar calls the “Moonshot” scenario highly uncertain—so the author recommends staying on the sidelines for now despite raw upside potential.

Analysis

The real market signal here is not a private valuation headline; it is how much additional AI spend investors are willing to capitalize into adjacent winners. In the next 1-3 months, that mostly supports the current semicap chain and data-center buildout names, because they monetize today’s capex, whereas the more speculative end uses heroic assumptions with very long payback periods. That makes the nearest-term upside more tangible for NVDA than for any “AI platform” story that depends on future customer acquisition and ecosystem lock-in.

The second-order risk is that the narrative can flip from “AI TAM expansion” to “capital intensity skepticism” very quickly if hyperscaler spend decelerates or margins fail to convert. If investors start to treat every large private AI/space valuation as proof of excess, the first public names to compress are the highest-duration software names, with PLTR more exposed than NVDA because it is priced more on future operating leverage than current hard demand. Any move in the private market itself is largely irrelevant unless it translates into disclosed procurement, launch contracts, or compute orders.

Contrarian take: the consensus is probably overestimating the value of vertical integration in moonshot businesses and underestimating how much of the economics still accrues to suppliers of picks-and-shovels. If in-space compute remains a distant option rather than a real program, the upside claim is a distraction; if it becomes real, the obvious winner may still be the GPU vendor and infrastructure stack, not the platform aspirant. The thesis is falsified if AI capex guidance rolls over, or if public-market multiples for semis fail to hold after the next earnings cycle.

Key Decisions for Investors

• Stay out of SPCX as a direct expression: the valuation discussion is too detached from observable cash flow and there is no clean public-market entry point. Use it only as a sentiment gauge for speculative AI optionality over the next 6-12 months.
• Buy NVDA on any post-headline weakness rather than strength; the stock benefits from actual capex today, while the 'future in-house GPU' narrative is too remote to impair near-term demand. Risk/reward is favorable as long as datacenter backlog and guide remain intact.
• Consider a tactical long NVDA / short PLTR pair on AI-exuberance rallies over the next 1-3 months. NVDA has more visible revenue conversion from the AI cycle; PLTR is more vulnerable if the market starts discounting long-duration AI promises.
• Set an alert on any earnings-season capex deceleration from hyperscalers; that would be the clearest falsifier for the broad AI upside thesis and a trigger to cut exposure to high-multiple AI beta.
• If you want a contrarian hedge, buy short-dated puts on a basket of the highest-duration AI software names rather than the semis, since the immediate reaction to 'AI bubble' rhetoric is more likely multiple compression than a collapse in actual chip demand.