Ahead of SpaceX’s plans to go public next week, Via Satellite surveyed leading space analysts about what we’ve learned from the company’s financial filings ahead of the IPO. The company’s S-1 filing, released in late May, gave the first real look into the company’s financials — including Starlink revenue, Starship and xAI spend, and debt levels, as well a view into the company’s long-term plans for AI.
Taking part in the roundtable are Nathan de Ruiter, partner and managing director of Novaspace; Caleb Henry, director of research at Quilty Space; Pravin Pradeep, senior consultant and program manager at Frost & Sullivan, Aerospace, Defence & Space; an Pierre Lionnet, research and managing director at Eurospace.
VIA SATELLITE: SpaceX has filed its initial documents for its IPO, making financial details public for the first time. What were some of the biggest takeaways from this information from your perspective?
de Ruiter: The IPO narrative goes well beyond rockets and satellite internet. SpaceX is positioning itself as a long-term infrastructure platform spanning space, global connectivity, and artificial intelligence. Notably, AI is framed as the largest component of its future market opportunity, suggesting the company wants investors to think of it less as a space company and more as a foundational technology and hyperscaler. The filings confirm that SpaceX is already operating at large commercial scale. The company generated roughly $18.7 billion in 2025 revenue, with Starlink emerging as the clear economic engine of the business. Starlink’s subscriber and revenue growth are outpacing what DirecTV achieved during its breakout years in the 1990s. At this point, Starlink can credibly be viewed as the most commercially successful space business ever built.
Henry: Confirmation of Starlink as the chief moneymaker (in line with Quilty expectations); the magnitude of xAI’s debt load (the clear driving force behind the IPO); and the CapEx spending on Starship/Super Heavy (+$15 billion through 2025, with more to come). Starlink began as a CapEx-heavy project itself, but has evolved into the cash generator for SpaceX’s next moonshots. Now SpaceX needs to go from one financial success story to a hat trick with launch and AI.
Pradeep: Starlink is the engine of the entire company. It generated $11.39 billion in 2025, making it the clear primary revenue driver with strong operating margins. Second, xAI is the giant hole in the balance sheet: the AI segment lost a whopping $6.355 billion in 2025, which is what’s driving SpaceX into the red overall, since the legacy Space and Connectivity segments are actually profitable. Third, the scale of ambition here is unlike any IPO in history. At an expected $1.675 trillion pre-money valuation, this IPO would generate more exit value than all VC-backed IPOs in the last decade combined.
Lionnet: The space and connectivity segments are both quite healthy and profitable, despite the heavy toll of Starship development costs ($3 billion in 2025). The AI segment is a drain on company finances, with huge losses, enormous CapEx, and growing debt. The huge amount of accumulated losses ($41 billion) tells us that current stockholders are far away from breaking even, unless they offload their stock. Particularly if, as expected, the AI business doesn’t create profits for quite a while.
VIA SATELLITE: If you would say that was one thing that really surprised you from these documents, what would it be and why?
Pradeep: The accumulated deficit. The most startling number in the S-1 is the accumulated deficit of $41.3 billion as of end of March, which is the running total of losses since inception. We knew SpaceX burned cash, but seeing that number in black and white, alongside a $4.3 billion net loss in just Q1 2026 alone, is a sobering reality check against all the hype.
de Ruiter: The most surprising element is how strongly the IPO is framed as an AI story, even though it is being presented under the SpaceX umbrella. The company’s proven businesses today are launch services and Starlink, but a significant portion of the valuation narrative is tied to more forward-looking opportunities including xAI, orbital compute, enterprise AI, and even space-based data centers. It raises a fundamental question: whether the same cost advantages and synergies that SpaceX has built through vertical integration in launch and satellite manufacturing can realistically be replicated in the far more competitive and capital-intensive AI market.
VIA SATELLITE: Starlink had over 10 million subscribers at the end of March this year. How do you see the potential of the connectivity market for SpaceX?
Henry: We estimate Starlink will end 2026 with around 17 million subscribers, driven by increases in network capacity, lower pricing, and access to more markets around the world. Importantly, Starlink is picking up meaningful growth in the enterprise sector, diversifying the business with higher ARPU customers.
Lionnet: It is clear that Starlink is already confronted to the law of decreasing margins in LEO. The S1 documents exhibits decreasing ARPU, and customer acquisition is slowing down. There will be a time when the most profitable segments of Starlink demand saturate (high income subscribers and enterprise and government business), and when adding subscribers may come at a very high cost. To grow its consumer customer base, Starlink must reach to increasingly densely populated areas, where the marginal efficiency of each additional Gbps deployed is decreasing fast. Without Starship enabling a launch cost reduction of an additional factor over Falcon, the Starlink infrastructure may not be able to profitably serve more than 25 million to 30 million households.
de Ruiter: Starlink’s 10 million subscribers have largely come from meeting demand that terrestrial networks failed to serve, rather than taking share from the legacy satellite broadband market, which was relatively small (peak at 2.9 million subs) before its entry. Its growth has been driven by the ability to deliver capacity at a disruptive cost base. Looking ahead, that advantage should strengthen further with Starship and next-generation satellites, enabling both better performance and lower pricing.
The key challenge is monetization. Monthly ARPU has declined from roughly $99 in 2023 to about $66 by early 2026, reflecting both expansion into lower-income geographies and more aggressive pricing in core markets. While that supports faster adoption, it puts pressure on margins. To offset this, SpaceX will need to continue scaling its subscriber base and driving down costs. SpaceX is also offering very aggressive prices, to lock customers as much as possible, before other constellations are entering service, which might explain a lower monthly ARPU.
VIA SATELLITE: SpaceX says it has the largest total addressable market (TAM) “in human history,” estimating a TAM of $370 billion in space, $1.6 trillion in connectivity, and $26.5 trillion in AI. What is your take on the addressable market, do you think this is realistic?
Lionnet: The TAM [total addressable market] narrative is completely off-track. Hyping the global connectivity market ($1.6 trillion — a figure that doesn’t even make consensus among telecommunications analysts) as TAM for Starlink despite knowing that 90% or more of that opportunity is out of reach due to the physics of LEO communications is misleading. Similarly the $370 billion opportunity for space is vastly irrelevant to SpaceX’s current business lines. But more than anything, the TAM is tied to tens of trillions of opportunity in AI. While I am not an expert in this field, I find the sheer value to be staggering to say the least, and I have a hard time buying it.
de Ruiter: The TAM is more of a narrative tool than a precise financial estimate. The space and connectivity markets are large but still understandable. The AI market number is much broader and less clearly explained. SpaceX’s “space” TAM should be viewed as a broad opportunity indicator rather than a directly addressable revenue pool. Around 60% of the estimate is linked to Novaspace’s concept of space-enabled solutions, which captures the wider economic value generated by services relying on space-based signals, including positioning, navigation, and timing. These capabilities enable large downstream sectors such as food delivery, ride-hailing, logistics, mapping, and finance. While SpaceX may provide critical infrastructure underpinning these markets, it would not capture the full value generated by application-layer platforms such as Uber or other downstream businesses. At the same time, the estimate is based on current market conditions and may understate the longer-term opportunity, as it does not fully reflect expected growth over the next decade from major developments such as Golden Dome, expanded orbital infrastructure, and lunar activities.
VIA SATELLITE: What do you think of the fact that the vast majority of TAM — 93% — is in AI, versus space and connectivity?
Pradeep: The AI number is where it gets speculative. TAM figures this large are ultimately unfalsifiable. No single company captures a majority of a $28 trillion market. AI applications [lead AI TAM] at $22.7 trillion, yet Space and Connectivity are the businesses generating actual revenue today. It tells you the story SpaceX is trying to tell investors: don’t value us as a rocket company, value us as an AI company. Whether that framing holds depends entirely on whether xAI can compete with OpenAI, Google, and Anthropic, which is far from certain.
de Ruiter: With around 93% of the TAM attributed to AI, the narrative relies heavily on AI to justify the valuation, as space and connectivity alone are unlikely to support it. AI is what introduces the perception of near-unlimited upside. The issue is that AI is also the least proven part of the SpaceX story. Investors are effectively being asked to price in a large opportunity where SpaceX does not yet have meaningful scale or a dominant position. SpaceX show signs of early traction, for example, large compute agreements like the multi-year Anthropic deal, but overall the AI business remains at an early-stage relative to its implied importance in the valuation.
More broadly, SpaceX’s future-market assumptions are highly speculative. The company is pointing to areas such as in-orbit manufacturing, point-to-point Earth travel, energy production on the Moon or Mars, and asteroid mining. These markets either do not yet exist commercially or remain at a very early stage, with unproven demand. They also face major regulatory, technical, safety and infrastructure barriers. That said, the underlying logic is directionally credible: for many of these markets to emerge, the sector will need a much cheaper and more scalable launch system such as Starship. In that sense, the investment narrative is compelling: fund Starship to unlock markets that are difficult to size and seize today but could become transformative tomorrow.
VIA SATELLITE: How do you review SpaceX’s decision to acquire xAI ahead of the IPO, considering that xAI generated the smallest portion of revenue and the largest loss from operations in 2025?
Pradeep: It’s a high-risk, high-reward move that’s hard to evaluate fairly this early. The AI segment generated $3.2 billion in revenue but posted a $6.355 billion operating loss in 2025. The strategic logic, combining Starlink’s global distribution with Grok’s AI capabilities and orbital compute, is coherent. But the timing raises governance questions. Combined with the voting structure, there’s real risk of decisions that benefit the broader Musk portfolio at the expense of SpaceX shareholders, and public investors have limited recourse given Musk’s 85% voting control.
Lionnet: I think that the decision to merge the AI business in SpaceX stems from the narrative of orbital data centers as the sole viable solution for massively scaling AI and compute. The acquisition of xAI is signaling the world that SpaceX do not plan to provide low cost access to orbit to anyone and would rather lose money with xAI than giving away this competitive advantage to a customer. It also signals the market that for SpaceX AI’s success is unequivocally linked to orbital compute. As the underdog in AI SpaceX probably needs to stand out from the competition and makes a bold bet. It will be interesting to see what the AI competition thinks of this, and whether they make a similar move in the future.
VIA SATELLITE: There has been a lot of talk recently about space data centers and this being the next big growth market for space companies. Are you believing the hype right now?
de Ruiter: Not in the near term. Space data centers will become interesting if launch costs fall dramatically and orbital solar power can be used efficiently. But there are several practical challenges to overcome, such as: maintenance, radiation, latency, regulation, thermal management, reliability and replacement cycles. Even if orbital data centers become real, it may take five to 10 years before they become a meaningful competitor to terrestrial data centers. Orbital edge computing can however represent an opportunity in the short term, with a credible need (reducing the traffic by downloading only useful information) and existing customers with a willingness to pay for those services.
Lionnet: This is a big hype but mostly in the space sector. Data center hyperscalers don’t seem to consider the opportunity seriously, even if a few have dipped a toe in the ODC ocean to feel the waters, I don’t see this as a seal of approval. The ODC business case hinges on three major technological breakthroughs: very high power to mass density of solar generators; compute hardware that withstands higher operating temperatures; and very low launch costs (below $200/kg). It will take a while before the technology stack enables viable ODC solutions for massive compute in orbit. The current energy demand bottleneck that is often used to justify ODCs is not enough of a driver in my opinion.
VIA SATELLITE: Do SpaceX’s financials justify a valuation of $1.75 trillion valuation? What are your thoughts on the valuation?
de Ruiter: Not on current financials alone. At $18.7 billion of revenue and a $4.9 billion net loss, the valuation cannot be justified using normal revenue or earnings multiples. The valuation depends on investors believing that SpaceX is credible to advance technologies at pace, and unlock the potential of several markets at once: launch, satellite broadband, direct-to-device connectivity, AI, orbital compute and eventually lunar or Martian infrastructure. In that sense, it is fundamentally a vision-driven valuation, with the TAM assumptions doing much of the heavy lifting. The key question for investors is whether one company can realistically dominate across so many emerging sectors and whether all of these markets will develop at the scale implied.
Pradeep: It’s a faith-based valuation at current financials. A lofty valuation of $1.5 trillion and potentially higher means early investors are paying a high premium at a time when financials aren’t all that strong. On 2025 revenue of $18.7 billion, you’re looking at roughly a 90x price-to-sales multiple. That’s only justifiable if you believe the AI and connectivity bets play out at enormous scale. If SpaceX were valued purely on its space and connectivity businesses, which are profitable, the number would look very different. It’s an AI valuation wearing a rocket suit.
Lionnet: Considering the statements on TAM and the current CapEx projections being driven mostly by AI I can only say that at that level of valuation the stock already factors in decades of massive AI/compute revenues, despite the fact that for the time being this business segment is losing billions with all current players. If SpaceX was only Space and Connectivity that are rather healthy and profitable, I’d say the company could be worth 10x to 20x EBIT like other successful wireless operators, somewhere in the $40 billion to $60 billion bracket. Pricing in the fact that the connectivity segment still has very comfortable margins could push the value a bit higher, but Starship remain a costly bet so far, and its success is a condition for growing the market in connectivity.
VIA SATELLITE: SpaceX detailed risk factors for future growth, including Starship development, regulatory concerns, dependence on Elon Musk, and the company’s level of debt, among others. What do you see as one of the most consequential risks to SpaceX’s growth?
Henry: The most consequential risk is xAI. SpaceX would be a much healthier company without xAI dragging down its financials. This could change in as little as 12 months, with xAI’s extraordinary cash burn offset by Anthropic, IPO proceeds, and new customers. Musk clearly sees xAI as a growth story, but for now it’s a drag on the business.
de Ruiter: The risk assessment is fair. The most consequential risk is execution, specifically, the need to deliver across multiple capital-intensive programs at the same time. Starship is the critical bottleneck: its timeline, cadence, and cost structure underpin nearly every part of the long-term story. Without Starship scaling as planned, it becomes much harder to expand Starlink capacity, roll out next-generation direct-to-device services, or enable future opportunities like orbital compute. In many ways, it all runs through that single chokepoint.
Pradeep: The Elon Musk dependency risk is the one that keeps institutional investors up at night. Musk holds 42% of economic interest but commands 85% of votes, meaning he can unilaterally approve mergers, acquisitions, capital raises, and executive compensation without public investor consent. His simultaneous leadership of Tesla, X, xAI, and now SpaceX as a public company creates a concentration risk with no parallel in public markets. Any significant distraction, whether political, legal, or reputational, lands directly on SpaceX’s stock price.
VIA SATELLITE: SpaceX will reportedly allocate a larger portion of its shares to retail investors than typical IPOs. What do you think of SpaceX’s strategy to engage retail investors?
Pradeep: It’s smart business with a dual motive. SpaceX’s CFO Bret Johnsen explicitly said “retail is going to be a critical part of this and a bigger part than any IPO in history,” citing their long-term support of Musk. The strategic logic is also product-driven. An investor in a stock is more likely to subscribe to Starlink or use X, meaning a large retail allocation can improve cash flows through more product users. That said, demand will likely far exceed supply, meaning most retail investors will receive only a fraction of what they request, if anything.
Lionnet: I think that Musk is aware of his very large fanbase and counts on them to prop-up the stock value until index buyers are forced into the market. Enough engagement from the retail base would keep the demand (and stock price) high for long enough. There is however, the risk that the Musk and space stock fanbase doesn’t have enough free money to buy $25 billion worth of SpaceX, some may liquidate part of their positions in Tesla or in existing similarly high-priced space stocks (such as Rocket Lab and AST SpaceMobile) to satisfy their crave for SpaceX. When such an outsized stock offering comes on the market the amount of new money it can draw to the marketplace is a critical aspect to consider. Will retail be able to afford SpaceX as expected?
VIA SATELLITE: If SpaceX is able to execute across some or many of these plans, will the company have unparalleled dominance in the space market?
de Ruiter: Yes, it already has. Starship would likely widen that gap further, reinforcing SpaceX’s cost and scale advantage. At the same time, it could also unlock opportunities for others, e.g. particularly players whose business cases depend on lower launch costs, or regional champions in other geographies that can leverage more affordable access to space to build competitive positions in their own markets. But the question is not whether the upside is large but whether SpaceX can execute without overextending itself financially and/ or operationally.
Henry: SpaceX already has unparalleled dominance in launch and in connectivity. Executing well on AI in space will likely give the company another pole position. That said, it’s still too early to tell. xAI was on a troubled course, and needs SpaceX to right the ship before it can talk about taking the lead.
Pradeep: Potentially yes, and uniquely so. SpaceX commanded more than 80% of global mass to orbit in 2025, and no competitor is close on launch costs. If Starlink captures meaningful broadband market share globally and xAI/orbital compute becomes a real infrastructure layer, SpaceX wouldn’t just dominate the space market. It would be critical infrastructure for the internet, AI, and defense simultaneously. The risk is that dominance in space doesn’t automatically translate to dominance in AI, where the competition is far more intense.
