OpenAI and SpaceX now represent the majority of institutional-grade venture secondary volume, a structural narrowing that forces family offices and fund managers to accept opacity or sit out the only liquid names in late-stage venture. The shift reflects exit scarcity, not quality discovery.
The venture secondary market historically distributed flow across 15 to 20 companies per quarter with observable price discovery. That base has collapsed. OpenAI's last tender valued the company at $300 billion. SpaceX cleared $350 billion in its December secondary round. Both trades moved $1.5 billion to $2 billion in single sessions, volume that once required a dozen separate portfolio events. The concentration is mechanical. No other private company offers comparable liquidity without triggering full exit waterfalls or hostile board dynamics. Allocators need somewhere to rotate capital while IPO windows stay closed. These two names absorb that need.
The concentration creates asymmetry family offices cannot model cleanly. Pricing in venture secondaries depends on information leakage from tender participants, employee chatter, and backwards-solved cap-table math. OpenAI runs tenders every six months with strict NDA clauses. SpaceX secondaries occur irregularly, often announced after settlement. The result is bid-ask spreads of 8% to 12% on $50 million blocks, wider than any publicly traded equity outside distressed situations. Fund managers buy because they must show exposure. Single family offices buy because their principal read the earnings call transcripts and decided the valuation lag versus Nvidia justifies the structure. Neither group has clean comps.
The data gap is not fixable with better reporting. Venture secondaries exist to delay the IPO moment, which means participants optimize for non-disclosure. OpenAI has no obligation to file financials. SpaceX treats revenue figures as classified unless a Starlink spin becomes unavoidable. Even the sophisticated allocators at Coatue or Tiger Global price these positions using revenue multiples borrowed from public SaaS comps, then apply a 25% to 35% illiquidity discount that has no empirical basis beyond tradition. The model works only if the exit multiple expands faster than the discount decays. That assumption held for ten years. It has not been tested in a sustained high-rate regime.
Family offices and fund allocators should monitor three follow-on developments. First, OpenAI's next tender round, expected in Q4 2025, will show whether the $300 billion valuation was participant-driven or guided by the company. Second, SpaceX's Starlink separation mechanics, likely formalized by early 2026, will establish whether the parent discount persists or compresses. Third, the SEC's renewed interest in private-market transparency rules, currently in comment period, could force disclosure within 18 months that collapses bid-ask spreads but also kills the anonymity these tenders depend on.
Venture secondaries have become a two-stock index with no reference rate and no exit calendar. The allocators buying in are not confused. They are choosing known opacity over the unknown duration of everything else in late-stage venture.