SpaceX is public. This happened four days ago.

June 12, Nasdaq, ticker SPCX, IPO price $135 per share. First-day close: $161.11, up 19.3%. Intraday market cap crossed $2 trillion. $75 billion raised — the largest IPO in history by roughly a factor of two over Saudi Aramco's $35.4 billion in 2019. SpaceX is now the sixth-largest publicly traded company in the US by market capitalization. SPCX was trading at $177.99 as of June 15, the day I'm writing this.

The part that matters for this audience: the Cursor acquisition clock is now running.

The deal structure, quick recap

SpaceX has the right to acquire Anysphere (Cursor) for $60 billion, approximately 30 days after the IPO — placing the expected close around July 12. Alternative in the deal: SpaceX pays $10 billion for collaborative work and stops there. A $10 billion breakup fee is attached.

SpaceX merged with xAI in February 2026 in a deal Elon Musk valued at roughly $1.25 trillion. xAI runs Grok. So the entity acquiring Cursor is, in operational terms, the same entity that builds Grok — and Cursor is currently one of the primary distribution channels where that model will be pushed.

67% of Fortune 500 companies use Cursor. That's the reach xAI is buying.

The Grok V9-Medium problem

Here's where the article gets uncomfortable.

Elon Musk confirmed on June 5 that Grok V9-Medium completed training — 1.5 trillion parameters, roughly three times the current production model. The model was trained on Cursor workflow data: real developer sessions, not synthetic code tasks. Public release was targeting mid-June.

It's June 16. V9-Medium has no published public benchmarks.

The "post-acquisition" column above is inference, not confirmed. xAI hasn't published a model routing plan for Cursor post-close. What is confirmed: xAI trained V9-Medium on Cursor data and is rolling Grok into Tesla and X as a distribution play. The financial logic for pushing Grok through Cursor is obvious.

The thing you can't verify yet: whether Grok V9-Medium is actually as good at coding as Claude Fable 5 (currently offline, US government export order) or GPT-5.5. No independent benchmark numbers exist for V9-Medium. Grok Build 0.1, a smaller dedicated coding model in API beta since May 29, is a separate product — a signal of direction, not a substitute for V9-Medium benchmarks.

Why Musk's timeline track record matters here

The mid-June target for V9-Medium has slipped. Four days into the second half of June, no release. I don't think V9-Medium is broken — training a 1.5T model on new data is a multi-month project, and "mid-June" was always aspirational. But it means the Cursor acquisition closes in about four weeks, and the model xAI intends to push through Cursor post-acquisition hasn't shipped yet, let alone been evaluated against the models it's replacing.

Or maybe it ships this week. I could be wrong about that. But even if it does, four weeks isn't much runway for builders to evaluate a 1.5T parameter model against their actual production workloads before it becomes the default routing in their IDE.

Source spread

• CNBC — SpaceX IPO closes at $161, jumping 19% after record debut [builder] Specific IPO close price, day-one performance, $75B raised.
• TechCrunch — SpaceX is working with Cursor and has an option to buy the startup for $60B [builder] Original deal structure reporting, confirmed by Bloomberg.
• TechTimes — Grok V9-Medium finishes training, rolling into Tesla fleet and X [hype] Training completion confirmation. No benchmark numbers published alongside.
• ChatForest — Grok V9-Medium Builder's Guide [builder] Best independent summary of V9-Medium specs and what's confirmed vs rumored.
• Big Hat Group — xAI Weekly: Grok Build 0.1 API [builder] Grok Build 0.1 public beta details, May 29 announcement.

Pros & cons

What's real:

• SpaceX's IPO was a genuine success. $75B raised at a $1.77T valuation. The company has the capital to execute the Cursor acquisition without financing risk. There's no scenario where this deal falls apart for financial reasons now.
• Grok Build 0.1 in API beta is a real signal. It's a dedicated coding model in the xAI lineup, different from V9-Medium, and it suggests xAI is building a coherent developer tooling strategy rather than just slapping Grok labels on things.
• If V9-Medium performs as advertised on coding tasks — and the Cursor workflow training data is a genuinely interesting differentiator — the Grok-in-Cursor story could be competitive. The bet that real developer workflow data beats synthetic training data is not obviously wrong.

What deserves a side-eye:

• V9-Medium has no published benchmarks. Mid-June came and went. The deal closes in four weeks. Builders who care about model routing in their IDE should want published numbers before the default changes, not after.
• xAI has a financial incentive to push Grok through Cursor. That's not a conspiracy — it's how business works. But it means user-configurable model routing may face quiet friction over time even if it stays officially available.
• Claude Fable 5 is currently offline (US DoC export order, June 12). Cursor's primary model routing is in an unusual state right now. The comparison baseline for V9-Medium's launch is murkier than it would normally be.

What builders need to know

• The Cursor acquisition closes around July 12, barring the $10B work-for-hire alternative. If you use Cursor in production, this is your planning horizon.
• Grok V9-Medium still has no public benchmarks as of June 16. Before the deal closes, publish your own comparison: your real workloads, Claude or GPT vs Grok Build 0.1, documented results. You want data when routing decisions get made.
• Grok Build 0.1 is in API beta now at grok-build-0.1. It's a smaller dedicated coding model, not V9-Medium. Worth running against your stack for directional signal.
• Watch for V9-Medium's public launch and independent evaluations on SWE-bench Verified and LiveCodeBench. If xAI doesn't publish against these standard benchmarks on launch day, ask why.
• User-configurable model routing in Cursor is currently available and should remain available post-acquisition. Build with that assumption, but verify it explicitly in the new terms when the acquisition closes.
• Anthropic's Fable 5 and Mythos 5 remain offline (US DoC export order). If you were relying on Fable 5 as your Cursor fallback, that baseline is currently unavailable. Plan around Claude Sonnet 4.6 or GPT-5.5 for now.

Further reading

• CNBC — SpaceX IPO closes at $161 — IPO day coverage
• CNBC — SpaceX stock jumps 20% in first full day of trading — June 15 update
• TechCrunch — SpaceX option to buy Cursor for $60B — original deal reporting
• TechTimes — Grok V9-Medium distribution flywheel — training completion coverage
• ChatForest — Grok V9-Medium builder's guide — independent spec summary
• xAI — Grok Build 0.1 API — the coding model currently in beta