The AI Build Revolt: Why 35% of Enterprises Have Already Replaced SaaS With Custom Code

On June 8, 2026, the roadshow begins for what may be the most consequential technology IPO since Google went public at $85 per share in 2004. SpaceX — formally Space Exploration Technologies Corp., now including Elon Musk's xAI subsidiary following their February 2026 all-stock merger — is targeting a $1.75 trillion valuation at its Nasdaq debut, scheduled for pricing on June 11 under the ticker SPCX. If successful, it will surpass Tesla to become the ninth most valuable public company in the world and represent the largest IPO by initial market capitalization in history.

But the story of SPCX is not primarily about rockets. It is about whether the AI valuation thesis — that foundation model infrastructure and distribution scale justify trillion-dollar company valuations while generating substantial operating losses — can survive contact with the quarterly earnings calls, SEC disclosure requirements, and institutional investor scrutiny of public markets. SpaceX's IPO is the first serious opportunity the market has had to answer that question at scale.

The combined SpaceX entity presents a fundamentally unusual investment proposition. Bundled inside SPCX are three distinct businesses with radically different financial profiles: a genuinely profitable satellite internet infrastructure business (Starlink), a proven and expanding launch services business, and an AI company (xAI) that generated $3.2 billion in revenue during 2025 while posting a $2.47 billion operating loss in Q1 2026 alone. Investors buying SPCX at $1.75 trillion are simultaneously purchasing Starlink's cash flows and betting that xAI's operating losses convert to competitive AI position on a timeline short enough to justify the current valuation.

What SpaceX Actually Is in 2026

The company that goes public this month is meaningfully different from the SpaceX that most investors conceptualize. The common mental model — Elon Musk's rocket company — reflects a business that has been substantially transformed by the Starlink build-out and the xAI merger.

The launch services business, which made SpaceX famous, remains one of the most technically impressive operations in the history of spaceflight. SpaceX has driven launch costs from an industry average of approximately $54,000 per kilogram to low Earth orbit to roughly $2,700 per kilogram — a 20x cost reduction that created the commercial satellite economy of the 2020s. Falcon 9's reusable first stage, now with hundreds of successful landings, is the engineering achievement that made this possible. But launch services, while high-margin and strategically important, are not the primary driver of SpaceX's current valuation.

Starlink is. The satellite internet business had 10.3 million subscribers across 164 countries in Q1 2026, generating $3.26 billion in quarterly revenue and $1.19 billion in quarterly operating profit — approximately $4.75 billion in annualized operating profit. That is a real infrastructure business with genuine margins in a market protected by the extraordinary capital cost of building the satellite constellation and operating the ground station network. Competitors are years behind. The fundamentals support a standalone valuation of $1.0 to $1.4 trillion depending on ARPU trajectory and terminal growth rate assumptions.

The xAI business, bundled in through the February 2026 merger, is the element that makes the combined entity's $1.75 trillion target most interesting to analyze — and most contested.

Starlink's Unit Economics: The Foundation and the Concern

Before examining xAI's numbers, Starlink's financial trajectory deserves careful attention because a trend buried in the aggregate revenue growth tells a story investors will need to evaluate.

Starlink's subscriber count has grown strongly: from approximately 3 million at the end of 2023 to 10.3 million today. That is a 240% growth in subscribers in roughly two and a half years. But ARPU — average revenue per subscriber per month — has declined from $99 in 2023 to $66 in Q1 2026, a 33% compression. Aggregate revenue has grown because subscriber growth has more than offset the ARPU decline, but the trajectory is important.

The ARPU compression reflects two dynamics. First, Starlink has expanded into consumer market segments in lower-income geographies where the consumer price point is significantly below the initial premium positioning. $99 per month for satellite internet is achievable in North American and Western European markets where the alternative is inadequate terrestrial broadband; it is not achievable in markets where per-capita income makes $99 a meaningful fraction of a monthly household budget.

Second, Starlink has introduced lower-priced tiers and expanded its service portfolio in ways that reduce blended ARPU even as they expand the addressable market. Enterprise and government contracts add high-ACV revenue at different unit economics than consumer subscriptions.

The critical question for Starlink's long-term valuation is whether subscriber count growth continues to outpace ARPU decline. If the current trajectory continues — ARPU declining 10-15% annually while subscribers grow 40-50% annually — aggregate revenue growth continues to look strong for several more years before the math reverses. But ARPU compression has already become a meaningful financial variable, and public market analysts will model it closely.

The xAI Integration: Operating Losses at Scale

The SpaceXAI segment's Q1 2026 results are the number that will drive the most discussion in SPCX's first earnings call as a public company. A $2.47 billion operating loss in a single quarter — annualizing to approximately $10 billion — is not exceptional in the context of AI infrastructure investment cycles. Amazon lost money for the first decade of its existence. OpenAI is estimated to be running multi-billion-dollar annual losses. Anthropic's explosive revenue growth to a reported $44 billion ARR in Q1 2026 came on top of substantial capital investment.

What is different about xAI's losses in the SPCX context is transparency. Private AI companies — Anthropic at $350 billion, OpenAI at $780 billion — are not required to explain their operating losses quarterly in a public filing reviewed by institutional analysts. SPCX will be. Every three months, the SpaceXAI segment's operating results will be published in a 10-Q, analyzed by equity research teams, incorporated into earnings models, and compared to the narrative Elon Musk uses on the earnings call. The AI infrastructure bet will be made in public.

The xAI distribution advantage through the X platform is genuine and material. Grok is accessible to X's 500+ million users, giving it a consumer distribution channel that no competing large language model can replicate without building a social platform from scratch. But distribution within X and commercial AI revenue are two different things. The enterprise API market — where OpenAI's developer ecosystem, Anthropic's enterprise contracts, and Google's Gemini API have established positions — is where AI revenue is most durable, and xAI is a meaningful step behind the leaders in enterprise AI adoption.

The Colossus data center in Memphis represents xAI's infrastructure bet: a large-scale GPU cluster that supports both internal model training and third-party compute sales. This asset is genuinely valuable — large-scale AI compute capacity is scarce and the barriers to building it are high. But Colossus competes in a capacity market that also includes Microsoft's Azure AI, Amazon Web Services, and Google Cloud, all of which have substantially larger installed bases and enterprise sales organizations.

The Valuation Math Laid Out

At $1.75 trillion, what is SPCX actually worth if you disaggregate the businesses and apply standard comparable multiples?

The intrinsic value analysis using comparable multiples suggests a range of $1.2 to $1.5 trillion before any conglomerate premium. The $1.75 trillion IPO target implies a 17% to 46% premium to this range, which is within normal bounds for highly anticipated IPOs with strong technical demand but leaves limited margin of safety for investors buying at or above the offering price.

The Nasdaq-100 Forcing Function

One of the most consequential mechanics in the SPCX listing is the automatic index inclusion that follows Nasdaq listing. Under Nasdaq's eligibility rules, SPCX qualifies for Nasdaq Composite inclusion immediately on listing and becomes eligible for Nasdaq-100 consideration after 15 trading days.

The Nasdaq-100 is not just an index. It is a mandatory purchase obligation for trillions of dollars in passively managed capital. Every fund tracking the Nasdaq-100 must buy SPCX shares proportional to its weight within the mandatory rebalancing window. QQQ alone manages approximately $300 billion in assets; dozens of other ETFs track the same index. At a $1.75 trillion market cap, SPCX's initial Nasdaq-100 weight would be approximately 3-4%, triggering an estimated $12 to $20 billion in forced passive buying from index funds alone, regardless of any individual manager's view on fundamental value.

This mechanic creates a temporary price floor that is structural rather than fundamental. The window between IPO pricing and full index inclusion is when momentum-driven retail trading and forced institutional buying are most active — and when SPCX's trading price is most likely to diverge from what a rigorous DCF analysis would imply. Smart money knows this. The short interest dynamics after index inclusion are complete will be worth watching closely.

The OpenAI IPO Shadow

SpaceX's SPCX debut does not happen in a vacuum. OpenAI — currently valued at approximately $780 billion in private markets following Amazon's commitment to invest $50 billion — has been publicly evaluated as a potential IPO candidate throughout 2026. The connection between SPCX's trading performance and OpenAI's IPO trajectory is direct and significant.

If SPCX sustains its IPO valuation and closes 2026 above its offering price, it establishes a public market reference transaction for AI-era trillion-dollar entities. It demonstrates that institutional investors will absorb massive AI infrastructure operating losses in exchange for long-term competitive position bets. It gives OpenAI's bankers a comparable transaction for their S-1 narrative. OpenAI could credibly price its IPO at or above $780 billion with that reference.

If SPCX declines materially after listing — as several highly anticipated tech IPOs have done when fundamentals caught up with narrative — OpenAI's IPO window narrows. The institutional investor base that was willing to absorb xAI's operating losses inside a combined entity with Starlink's cash flows will be far less willing to absorb OpenAI's operating losses in a pure-play AI company with no infrastructure cash flow offset.

The AI venture capital cycle of 2023-2026 has created private market AI valuations that have never been tested by public market disclosure and accountability. SPCX is the first significant experiment in whether those valuations survive contact with reality.

What the Public Market Test Will Actually Measure

The next 12 months of SPCX trading will measure several things simultaneously, and the results will reshape AI company valuations across the industry.

It will measure whether institutional investors in public markets, constrained by quarterly accountability and portfolio mark-to-market requirements, apply the same valuation logic to AI infrastructure businesses that private market investors have applied in the absence of these constraints. Private investors can hold through multiple years of losses with patience; public investors face quarterly redemption pressure and benchmark comparison that creates selling pressure at different points in the loss curve.

It will measure whether xAI's operating loss trajectory — $2.47 billion in Q1 2026 alone — inflects toward breakeven on the timeline that the $1.75 trillion valuation implies. The market is pricing in a specific AI business trajectory. Quarterly disclosures will reveal whether that trajectory is on track.

And it will measure whether Starlink's ARPU stabilizes or continues to compress as subscriber growth expands into lower-income market segments. Starlink's fundamental value as an infrastructure business depends heavily on this single variable in the long-term DCF model.

The SpaceX IPO is not a binary bet on rockets or AI. It is a simultaneous bet on three distinct businesses with different risk profiles, different competitive dynamics, and different timelines to return. Understanding which business you are betting on when you buy SPCX — and whether the current valuation appropriately prices each — is the analytical challenge that will separate thoughtful investors from those simply riding index inclusion mechanics.

Takeaway: SpaceX's $1.75 trillion IPO is AI's first serious public market test — a real price discovery event for AI infrastructure company valuations that have been set in private markets without the accountability of quarterly disclosure. The Starlink business has genuine fundamentals that support a $1.0 to $1.4 trillion standalone valuation. The xAI business is running $10 billion in annualized operating losses while competing in an enterprise AI market where OpenAI, Anthropic, and Google have more established positions. The Nasdaq-100 inclusion mechanic will create $12 to $20 billion in forced passive buying that supports near-term SPCX price performance independent of fundamentals. How SPCX trades in its first twelve months will set the reference price for every AI company IPO that follows — including OpenAI's — and will either validate or materially complicate the private market AI valuation thesis that drove $242 billion in AI venture capital in Q1 2026 alone.