A Hong Kong Firm Signs Up to Put $500m of Pre-IPO AI Shares Onchain

L2 Capital, a Hong Kong investment firm, has agreed to tokenize $500m of shares in privately held artificial intelligence companies through Newton Labs, according to reports from The Defiant and Crypto Briefing. L2 Capital is the first announced partner on the platform, which Newton Labs is still building.

Pre-IPO shares are stakes in companies that have not listed. They change hands today through secondary brokers, special purpose vehicles and forward contracts, with settlement measured in weeks and a buyer list that is short because every transfer has to be checked against the company's own share transfer rules. That friction, not a shortage of willing buyers, is what keeps the market narrow.

Compliance inside the token

The approach described is to encode the transfer restrictions into the token itself, so that each proposed transfer is validated against those rules before it settles rather than cleared by paperwork afterwards. In practice this means the token can refuse to move: if the receiving wallet is not on the permitted list, or a lockup has not expired, the settlement fails at the protocol level.

This is the design most serious private-asset tokenization has converged on, and it is the opposite of what makes a token liquid. A share that can only move to pre-approved holders does not trade freely on an open exchange. What it can do is settle faster between parties who are already eligible, and be used as a record that both sides and the issuer agree on.

$500m, and the sequence it sits in

The figure is a commitment to a platform that is not yet live, which means the number to watch is not $500m but whatever is actually issued first. Newton Labs has one partner. The size of that partner's book is a statement of intent, not an onchain balance.

The appetite behind it is easier to explain than the mechanics. Private AI companies have absorbed enormous amounts of capital at valuations that only mark when a new round prices, and investors holding those positions have limited ways to realise or hedge them. The same week, The Defiant reported that MNX launched futures referencing AI laboratory IPOs and revenue, contracts that give leveraged exposure rather than shares, with settlement tied to IPO filings, private-market valuations or revenue estimates. Two different answers to the same problem: one tries to move the share, the other tries to avoid needing it.

What would make this real

The test for any permissioned tokenization platform is whether a transfer that should fail actually fails, and whether one that should succeed settles without a human in the loop. Neither can be observed until the platform runs. Until then, what exists is a Hong Kong firm, a dollar figure and a design.

Nothing in either report says which AI companies are involved, or when issuance begins.