CFRA’s Keith Snyder says 71% of SpaceX value hinges on AI, not rockets

SpaceX’s IPO made headlines for being historic. Keith Snyder at CFRA says the hard part is deciding what, exactly, that $2 trillion market value is underwriting. His answer is blunt: he estimates xAI accounts for 71% of SpaceX’s valuation, and he can only square SpaceX’s current market cap by assuming “almost comical growth for AI,” even while acknowledging he still comes up short.

Snyder’s concern starts with SpaceX’s own S-1 framing. The filing effectively invites investors to dream big on AI, claiming 93% of its total addressable market would be “in AI,” a number Fortune reports as roughly equivalent to U.S. annual GDP, or $26.5 trillion. In the same S-1, SpaceX says it has identified “the largest TAM in human history,” and quantifies $28.5 trillion in total quantifiable TAM, allocating $26.5 trillion for AI. That allocation is broken out across $2.4 trillion in AI infrastructure, $760 billion in consumer subscriptions, $600 billion in digital advertising, and $22.7 trillion in enterprise applications.

Here is the tension. The rocket-and-satellite side of the business is described in the source as profitable and dominant, and Starlink has millions of paying subscribers globally. That matters because it gives SpaceX a real revenue and cash-flow base, not just a moonshot. But Snyder’s point, as Fortune lays it out, is about what it takes to justify the total market cap once you treat the AI story as a central pillar rather than an optional upside. In his view, most investors likely underweighted the AI assumptions embedded in those S-1 projections.

So what does the real-world data say about the AI revenue engine? According to the source, out of 117 million people who currently interact with Grok via its freemium model, only 1.6% pay for the premium tier. Fortune specifies paid subscribers as 1.9 million SuperGrok, SuperGrok Heavy, and SuperGrok Lite. Snyder, in comments relayed by Fortune, ties that relatively low conversion to the broader competitive landscape: other AI providers can deliver “everything better right now,” which reduces the incentive to sign up specifically for Grok when alternatives have more sophisticated models and positioning.

The source sharpens that competitive comparison with additional adoption and monetization context. New data from fintech Ramp’s AI adoption tracker puts xAI at just 3% business adoption, contrasted with Anthropic and OpenAI at 40%. The business adoption angle matters because enterprise buyers generally do not pay for novelty. They pay for stickiness: a workflow that becomes hard to replace. Kyle Poyar, a former operating partner at OpenView, is quoted by Fortune explaining the conversion mechanism in plain terms: if users are not adopting Grok on a weekly, ideally daily basis, and for sticky high-value use cases, they are likely never going to pay for Grok.

Poyar also adds a playbook-level monetization lens. He describes how AI companies increasingly treat token costs for free users as a marketing expense to draw in paid customers, rather than spending money on ads. He compares the approach by Anthropic and OpenAI to spending compute for users, which, he says, should accomplish the same purpose: attracting paid customers and, ideally, enterprise ones. Importantly, Poyar’s assessment is not universal praise for the strategy itself. He says, “I do think that strategy is working for Anthropic,” but adds, “It’s not clear to me that that’s working as well for xAI.”

Snyder’s critique then widens to a strategic product concern: model maturity and infrastructure leverage. He says Grok is still early compared with other AI models, and he calls out the risk of being framed as a “glorified Amazon Web Services” that rents out excess computing power to Google and Anthropic. The implication in the source is straightforward: if xAI cannot evolve its own model complexity and sophistication to compete, it may struggle to turn freemium attention into premium revenue at scale. Snyder also points to what the 1.6% number signals: other providers can offer better options now, which reduces the incentive to upgrade and reduces enterprise momentum.

There are also second-order economics to consider, and the source gives two academic or measurement perspectives that matter for boards and CFOs. Vineet Kumar, a professor at Purdue University’s Mitch Daniels School of Business known for research on freemium models, is quoted saying there is a tradeoff: if a firm is more generous with its free offering, more users sign up. But if it is too generous, users have less reason to upgrade. Then Evan Bailyn, whose company First Page Sage tracks and publishes B2B conversion rates across industries, provides a cultural analogy that is surprisingly practical: unlike Spotify, which enjoys high conversion because it is integral to how people exercise and socialize and relax, Grok is more of a “nice to have” add-on to X with no real incentive to be premium. Bailyn does not expect Grok to become a front-runner in the enterprise AI race, especially since the source says OpenAI, Google Gemini, and Anthropic already hold most of the enterprise market share.

For executives and investors, the underlying stakes are bigger than one AI product’s conversion rate. The source is really about how capital markets assign probabilities to AI growth when a company’s valuation depends on those probabilities. SpaceX is pitching an “AI everywhere” TAM story in the S-1, but the adoption and monetization signals highlighted by Fortune suggest that translating user interest into paid, sticky, enterprise usage may be harder than the headline TAM math implies. If that gap persists, it does not just challenge xAI’s revenue projections. It can also change how boards and investors discount the AI portion of any “platform + distribution + monetization” strategy, and how aggressively they should underwrite AI TAM expansion when the conversion funnel does not yet match the story.