SpaceX's S-1 Shoots the Moon

May 20, 2026 | Authors: Alex Gove & Connie Loizos

---

1. Key Themes

Theme 1: The AI IPO Wave Is Finally Breaking

After years of private-market dominance, the two most consequential AI companies are moving toward public markets simultaneously — and the scale is historic.

"SpaceX's expected IPO could become the largest in U.S. history, potentially raising around $75 billion and dwarfing previous record-holders including Alibaba's $22 billion 2014 debut while also helping revive an IPO market that has remained sluggish since 2021."

"OpenAI is reportedly preparing to confidentially file IPO paperwork within days, potentially setting up a public debut as early as September after clearing a major legal obstacle in Elon Musk's lawsuit."

---

Theme 2: AI Infrastructure Is Now a Revenue Business — Not Just a Cost Center

The Anthropic-xAI compute deal reveals that AI infrastructure is generating enormous, contracted cash flows. This reframes data centers from capital expenditures into revenue-generating assets with long-term enterprise contracts.

"Anthropic has agreed to pay xAI roughly $1.25 billion per month through 2029 for access to compute capacity at xAI's Colossus data center in Tennessee, a deal that could generate more than $40 billion in revenue for xAI."

---

Theme 3: Nvidia Is Becoming the AI Ecosystem's Venture Arm and Strategic Consolidator

Nvidia's dealmaking activity has quietly surpassed the scale of most institutional VC firms, and it's not coming from its formal VC arm — it's coming from business development, tying customers and suppliers directly to the Nvidia stack.

"Nvidia has committed roughly $90 billion to investments, partnerships, and dealmaking across more than 145 AI-related companies over the past 16 months, with much of the activity reportedly driven not by its NVentures VC arm but by Nvidia's business development group as the company ties customers, suppliers, and startups more closely to its chips and software ecosystem."

"Nvidia posted another record quarter with $81.6 billion in revenue and disclosed that its holdings in privately held startups nearly doubled to $43 billion over the past quarter."

---

Theme 4: AI Is Disrupting Legacy Software From the Inside Out

Traditional software incumbents are not just pivoting to AI — they are being forced into painful restructurings as AI-native competitors threaten to make their core products obsolete. This is not a gentle transition.

"Intuit is laying off roughly 3,000 employees, or 17% of its workforce, as the maker of TurboTax, QuickBooks, and Credit Karma restructures to focus more aggressively on AI amid broader fears that traditional software companies could be disrupted by AI-native competitors."

"Meta kicked off one of the largest layoffs in its history today, cutting roughly 8,000 employees — or 10% of its workforce."

---

Theme 5: OpenAI Is Playing an Aggressive Platform Land-Grab at the Startup Layer

By offering AI tokens in exchange for equity across an entire YC cohort, OpenAI is systematically embedding itself into the cap tables and infrastructure of the next generation of startups — before they have valuations.

"Altman offered $2 million worth of OpenAI tokens to every startup in the current class in exchange for equity in the startup."

"The deal will be offered as an 'uncapped SAFE,' meaning, 'it will convert in the next priced round, which is typically the Series A.'"

---

2. Contrarian Perspectives

Contrarian 1: SpaceX's IPO May Be More of an AI Story Than a Space Story

The S-1 framing reveals that SpaceX is positioning itself not as a launch company, but as an AI infrastructure empire — complete with orbital data centers and billions in losses tied to a separate AI venture. Investors buying the SpaceX IPO may unknowingly be buying an xAI proxy with attached governance risk.

"SpaceX files its S-1, revealing an Elon Musk empire increasingly built around AI — with billions in losses tied to xAI, plans for orbital data centers, and governance structures that leave Musk holding more than 85% of voting power."

This is a significant governance red flag that public market investors have historically penalized — yet the expected $75 billion raise suggests the market may be willing to overlook it entirely given the hype cycle.

---

Contrarian 2: The Anthropic-xAI Compute Deal Is a Competitive Red Flag for Anthropic, Not a Win

On the surface, Anthropic securing compute capacity looks like smart infrastructure planning. But paying a direct competitor ($1.25B/month to Elon Musk's xAI) to keep your AI running is a dangerous dependency — and an enormous capital transfer to a rival.

"Anthropic has agreed to pay xAI roughly $1.25 billion per month through 2029 for access to compute capacity at xAI's Colossus data center in Tennessee, a deal that could generate more than $40 billion in revenue for xAI."

The consensus read is "big compute deal." The contrarian read: Anthropic is funding xAI's growth while remaining vulnerable to supply disruption from its own competitor.

---

Contrarian 3: OpenAI's YC Token Deal Is a Valuation Trap for Founders

The "mic drop" framing of Sam Altman's offer obscures a structurally unfavorable deal for founders. An uncapped SAFE means OpenAI's equity conversion happens at whatever valuation the Series A sets — giving OpenAI maximum upside with no price ceiling, while founders receive tokens whose future value is uncertain.

"As for how much equity each startup can expect to give up, that can't be determined at the time it signs the deal. It will depend on how much the startup is worth when it raises its first priced round."

"It will convert in the next priced round, which is typically the Series A."

Founders receive $2M in tokens of uncertain value; OpenAI receives uncapped equity in 169 startups. The headline is generous; the terms deserve scrutiny.

---

3. Companies Identified

---

4. People Identified

---

5. Operating Insights

Insight 1: Use Token-for-Equity Deals Strategically — But Negotiate the Cap

OpenAI's YC deal signals that AI platform companies will increasingly offer compute credits as currency for equity. As a founder, this creates a new negotiating dynamic: compute credits have real operational value, but an uncapped SAFE is among the most dilutive structures available.

"As for how much equity each startup can expect to give up, that can't be determined at the time it signs the deal. It will depend on how much the startup is worth when it raises its first priced round." Tactic: If offered token-for-equity deals, push for a valuation cap or discount rate to limit downside dilution at Series A.

---

Insight 2: AI Is Now a Restructuring Mandate, Not a Feature Roadmap

Companies like Intuit are not adding AI features — they are reorganizing their entire workforce around AI adoption under competitive duress. For operators, this signals that AI transformation requires organizational redesign, not just tooling upgrades.

"Intuit is laying off roughly 3,000 employees, or 17% of its workforce, as the maker of TurboTax, QuickBooks, and Credit Karma restructures to focus more aggressively on AI." Tactic: Map your org chart against AI-replaceable functions now, before competitors force the issue at scale.

---

Insight 3: Infrastructure Moats Are Being Built Through Strategic Investment, Not Just R&D

Nvidia's $90B in dealmaking is primarily a business development function, not traditional VC. It ties customers, partners, and startups to the Nvidia ecosystem through financial relationships — a powerful and underappreciated moat-building playbook.

"Much of the activity reportedly driven not by its NVentures VC arm but by Nvidia's business development group as the company ties customers, suppliers, and startups more closely to its chips and software ecosystem." Tactic: For enterprise infrastructure companies, consider strategic minority investments in key customers or partners as a retention and lock-in mechanism — not just a financial return strategy.

---

6. Overlooked Insights

Overlooked Insight 1: The $200B AI Agent CPU Market Is a New Investment Frontier

Jensen Huang's claim that AI agents require a purpose-built CPU category — not just GPUs — represents a potentially significant platform shift. If billions of AI agents come online, the inference and orchestration layer may require fundamentally different silicon than current AI workloads.

"Nvidia has discovered a 'brand new' $200 billion market centered on CPUs designed specifically for AI agents, arguing that the company's new Vera chip opens a massive new business beyond Nvidia's traditional dominance in GPUs as billions of AI agents eventually come online." This is barely a footnote in the article but could be one of the most consequential hardware investment themes of the next five years.

---

Overlooked Insight 2: OpenAI's Reasoning Model Has Produced Original Mathematical Proofs

In a single line, the article notes that OpenAI's reasoning model disproved an unsolved conjecture posed by mathematician Paul Erdős in 1946. This is a qualitative capability leap — original mathematical discovery — that has enormous implications for scientific research, drug discovery, and any domain requiring novel reasoning at scale.

"OpenAI says one of its new reasoning models has produced an original proof disproving an unsolved geometry conjecture first posed by mathematician Paul Erdős in 1946." The buried lede: AI is no longer just retrieving or summarizing knowledge — it is generating it.