Private markets are full of inefficiencies.
Most people just don’t see them.
I spend my days looking at late-stage tech deals, secondary liquidity and allocation opportunities.
This account is where I’ll write about 👇
- private market inefficiencies
- secondary market arbitrage
- mispriced late-stage tech
- liquidity-driven opportunities
- SPVs and allocation access
- AI company secondaries
- signals before IPOs and M&A
NFA.
A single private company raising $30B in one round is roughly the entire annual mid- and late-stage VC market a few years ago (per Forge data).
That capital flow used to require public markets. It doesn't anymore.
The plumbing for backing the largest companies in the world has quietly migrated to a handful of private-market desks.
"Anthropic trades at $1T on secondary markets" reportedly means: a small number of clips, on one platform, with limited supply, that the company may not even recognize.
The actual market is the next primary round. Everything else is a screenshot.
"AI is in a bubble."
"AI is the most under-priced trade of the decade."
Both camps are looking at the same screen. The difference is whether you're pricing the technology or pricing the capital structure financing it.
The technology might be early. The financing layer is late.
Three months ago Forge was pricing OpenAI above Anthropic. Today it's the opposite, by ~$120B.
Same two companies. Same compute story. Same two sets of investors.
What changed wasn't the AI race. It was the supply: one side had sellers, the other didn't. Secondary prices reveal pressure, not consensus.
In Q1 2026, secondary buyers reportedly bid up Anthropic clips to $960B inside hours (per Rainmaker, Business Insider).
In the same quarter, retail private credit funds were reportedly gating redemptions and the public BDC index hit a ~17% discount to NAV.
Same word, "private." Two opposite liquidity conditions.
Anthropic at $9B ARR in late 2025. Reportedly $30B ARR by March 2026. Reportedly raising at $900B in May.
The cycle from product launch to civilizational-scale capital, fully inside the private market, with no public investor allowed in the trade.
This is what "private for longer" actually looks like at the limit.
Liquidity is a feature of structure, not of price.
You can have a $1T valuation and no way to sell. You can also have a $20B loan and no way to mark it.
Tenders are the new IPOs.
Continuation funds are the new exits.
Secondaries are the new pricing layer.
Private credit is the new high-yield.
The whole capital-formation stack rebuilt itself outside the public markets, and most people are still reading public quarterly reports.
Signs the private capital cycle is later than it feels:
1/ Goldman flags five danger signals echoing dot-com
2/ Public BDC index trading at ~17% discount to NAV (CAIA)
3/ Private credit shadow default rate ~6% vs stated ~2%
4/ AI ARR doubling quarter-on-quarter
5/ Secondary valuations of single companies running 2x what bankers expect at IPO
None of these alone is the answer. All of them together is the question.
Signs late-stage secondaries are consolidating, not just growing:
1/ tenders described by senior practitioners as "the new IPOs"
2/ time-to-IPO median now past 8 years
3/ US VC LPs sitting on ~$197B of negative cumulative cash flow since 2022
4/ banks building private capital advisory desks
5/ regulators requiring independent valuations (SEC 2023 rule)
(figures per Arcanis, May 2026)
Signs a "$1T secondary valuation" headline deserves an asterisk:
1/ company controls transfer recognition
2/ only a handful of shares actually changed hands
3/ no employee tender open
4/ no recent primary at that level
5/ buyer pool is one platform deep
Headline price is not clearing price.
The bull case: AI is rebuilding the economy.
The bear case: private credit is funding it with the wrong instruments.
Man Group flagged it last week. Data-center debt is collateralized like real estate but depreciates like smartphones. The first default wave likely lands in 2027 to 2028, when initial leases renew.
Both cases can be true.
Late-stage private secondaries quietly became a $200B+ asset class with no agreed-upon screen.
That is a structural opportunity the size of an entire generation of financial infrastructure.
How a $1T secondary print and a blocked transfer can coexist, explained simply:
A platform quote is a price someone is willing to pay. A transfer recognition is the company agreeing the buyer is now on the cap table.
If the second one is missing, the first one is just a number on a screen.
Two numbers from the same week:
Anthropic reportedly raising at $900B. Lincoln International's database shows 6.4% of private credit loans now carry "bad PIK" (interest deferred mid-loan), up from 2.5% in Q4 2021 (per CAIA, May 2026).
Same capital cycle. Different ends of it.
Public equities got Bloomberg in the 1980s because everyone needed the same screen to argue about price.
Private markets are arriving at the same problem. Multiple platforms, no shared price, pre-2022 vintages still marked 30 to 60% off.
The next decade in late-stage secondaries will be about who hosts the price, not who finds the deal.
The Anthropic news the AI press is missing: the company has reportedly moved to block Forge, Hiive and Sydecar from offering its shares, and said unapproved transfers won't be recognized on the books.
Translation: the secondary market for Anthropic is now whatever Anthropic says it is.