It was the kind of Tuesday that crypto loves to interrupt. Within 72 hours, a notification thread rippled through Telegram VIP channels: Gate, the exchange that walked away from the 2022 contagion with its balance sheet still breathing, is opening Phase 3 of its Pre-IPO token program. This time, the target is Moonshot AI, the Chinese AI unicorn behind Kimi, packaged as a "Mirror Note" at an implied $50 billion valuation, with a subscription band of $105 to $115 per share.
The underwriting fee is 5%, payable before your position exists. The dedicated secondary market, launching roughly a month after distribution, will charge takers 1.5% plus another 1% for the privilege of trading inside Gate's walled garden. The minimum ticket is $10,000. And if the IPO somehow fails, Gate will settle "according to applicable rules," which is the kind of language that makes a lawyer smile and an investor wince.
I've spent enough years inside financial engineering to know that fee structures are the first confession a product makes. Back during DeFi Summer, I audited more than 150 Uniswap V2 liquidity pools and found a critical edge case in slippage calculation that put $2 million of user funds at risk. That experience taught me something that has never stopped being true: liquidity isn't a feature; it's a relationship. And this product is a specific kind of relationship, one where the counterparty controls the mirror, the market, and the terms of reflection.
Context: The Product in Its Ecosystem
Let's start with what is actually being offered, because the marketing language is doing enormous heavy lifting. A Mirror Note is a synthetic instrument designed to track the value of an underlying company's shares without transferring legal ownership of those shares. In traditional finance, this structure closely resembles a contingent payout note, or a synthetic long exposure contract. The user is not buying equity in Moonshot AI. The user is buying a contractual claim against Gate, whose value Gate determines based on the performance of the referenced asset.
This is not Moonshot AI's fundraising round. The company did not issue these tokens, has not publicly endorsed this product, and its legal team is likely already drafting questions about brand use. The origin of the underlying shares that stand behind the reference โ if any exist โ is not disclosed. Neither is the custody arrangement. Neither is the independent audit trail. We have, in essence, a centralized platform issuing a tokenized claim referencing a private company's value, sold to retail users under platform-defined terms, with platform-controlled settlement.
The "Kimi" branding does what branding always does: it short-circuits careful analysis by invoking a beloved product. Kimi is a genuinely impressive generation of Chinese large language model applications, and Moonshot AI's founding team carries real technical credibility from the Tsinghua lineage. As someone who spends my days studying how AI intersects with cryptography, I hold no skepticism toward their engineering. But none of that technical competence transfers to the token being sold. The engineers building Moonshot AI are not building Gate's Mirror Note. Their execution excellence matters for the underlying business; it says nothing about whether the synthetic instrument will settle as advertised.
What we know with confidence from the announcement: Gate calls this Phase 3, implying at least two earlier phases of Pre-IPO issuance have been completed without drama. Gate claims a "comprehensive ecosystem" spanning Pre-IPO access, IPO participation, stock trading, and gStocks tokenized securities. Gate says it serves over 58 million users globally and maintains a 100% reserve proof claim. All of that is context worth registering.
What we don't know is far more important. The identity of the custodian holding the reference shares. The legal terms governing the note issuance. The settlement path in a bankruptcy scenario. The precise conditions under which the token can be cancelled. The subscription materials' reference to "applicable rules" for refunds is a black box, bolted shut, with no key offered to the buyer.
Core I: The Fee Labyrinth
Load the numbers once, slowly, and the fee structure reads less like a product feature and more like a stress test of user tolerance. The subscription band sits between $105 and $115. Take the midpoint: $110. The 5% underwriting fee is charged at subscription, which means you begin at negative 4.76%. To reach breakeven, Moonshot AI's IPO must value the company at least 5.26% above the mid-point price. To be clear: this is the threshold purely for getting your initial capital back. All price appreciation below that line belongs to Gate, not to you.
Now layer in the exit costs. The dedicated secondary market charges 0.5% for makers and 1.5% for takers. On top of that sits a separate 1% "dedicated market trading fee." A taker selling a position therefore pays 2.5% in combined friction. Adding the 5% underwriting cost and the terminal exit friction together, your all-in measurable cost reaches roughly 7.5%. The underlying equity must appreciate 8.1% before you book a single dollar of net profit.
Then comes the performance fee: 20% of "excess returns," a phrase never clearly defined. Is the benchmark the subscription price? The IPO listing price? The secondary market price at time of exit? The ambiguity is not accidental. The 20% carry is a claim against your upside with variable geometry, and the absence of a defined reference point means the platform retains maximum interpretive latitude when settlement day arrives.
Compare this with the alternatives that already exist.
In the traditional pre-IPO market, platforms like Forge Global and EquityZen charge integrated fees that typically total under 5% per transaction, and they settle in actual equity. Their investors receive actual shares, actual transfer rights, and a legal claim against the company. On Gate, you are paying more in fees to receive less legal substance: no direct share ownership, no voting rights, no enforceable claim against Moonshot AI. And you cannot exit into a competitive open market. You must sell into the exchange's dedicated venue, to other users of that same exchange, at prices formed by however thin the order book is.
Ondo Finance's tokenized Treasury products charge fees that are a fraction of Gate's structure, and they provide exposure to liquid, yield-bearing, regulated collateral. Backed Finance issues tokenized versions of publicly listed equities with near-zero marginal trading costs. Even FTX, in its doomed attempt at tokenized stocks, managed 0% trading fees plus 0.1% settlement fees. The Gate KIMI note manages to be simultaneously more expensive and more opaque than the failed FTX model. That is not an accident. It is a signature.
Core II: The Financing Desk Angle
Let's talk about the yield subsidy, because the 3.8% APR tells a story that the marketing slides omit. Users who subscribe with GUSD are promised 3.8% APR from U.S. Treasury yields, distributed daily, while their capital waits for allocation. Users who subscribe with USDT receive the same 3.8% "subsidy," calculated from hourly snapshots. On the surface, this looks generous. Beneath the surface, it is a retention yield.
Gate collects idle USDT and GUSD deposits across its user base. That stablecoin inventory earns yield for the platform's treasury โ in short-duration Treasuries, money market funds, or whatever instrument the treasury desk chooses. The current short-term U.S. Treasury yield environment is well above 3.8% for institutional holders. Matching a portion of that yield back to unallocated subscribers is a capital retention strategy disguised as a benefit.
The design becomes clearer when you consider the minimum subscription of 10,000 USDT or GUSD. That threshold is not casual. It targets users with meaningful stablecoin balances, users who might otherwise be tempted to move capital elsewhere in a sideways market. The pre-IPO narrative โ AI, China, unicorn, early access โ serves as the emotional hook that keeps that capital parked. Even if only a fraction of subscribers receive allocation, the rest continue accruing 3.8% "while you wait." The product quietly functions as a deposit retention mechanism dressed in venture capital clothing.
And the VIP airdrop program sweetens the bait further. It is designed to attract high-net-worth participants, the exact segment of users whose balances matter most to an exchange's treasury operations. The psychology is classical: the airdrop makes participation feel like a reward, when the fee structure has already priced the reward into the cost of entry.
Core III: What the Mirror Doesn't Show
The deeper question is what happens when the mirror breaks. Mirror Notes are only as solid as the underlying share arrangement, and we know almost nothing about it.
Consider the scenario that private equity lawyers think about before anyone else: the right of first refusal. Moonshot AI's existing shareholders almost certainly carry ROFR provisions in their agreements. If Gate purchased or borrowed Moonshot AI shares to hedge the Notes, those shares may be subject to rights that force their transfer to existing investors at original purchase prices. A triggered ROFR could strip the asset backing from the Notes, leaving Gate to settle "according to applicable rules." In legal drafting, that phrase is not a protection. It is a reservation of discretion.
There is also the question of valuation integrity. The $50 billion implied valuation sits in a band that private market narrative might justify but public market discipline might reject. At $50 billion, Moonshot AI is priced for enormous market capture in a sector where revenue multiples remain contested territory. Yet the token's price discovery will be conducted inside Gate's closed market, without the participation of independent institutional market makers, without the discipline of a public order book, and without any outside arbitrage mechanism that could correct divergence from private market reality.
Here I must draw on a thesis I have argued since my days on the financial engineering desk: orderbook DEXs will never truly rival CEXs because professional market makers refuse to put quotes on-chain when latency equals getting front-run. But this product inverts the problem in a disturbing way. On Gate's dedicated market, there are no external market makers at all. There is no cross-venue arbitrage. There is no transparent feed that can validate the quoted price. The "market" is a single room with one window, and Gate controls the shutters. Without disclosure of order book depth, bid-ask spread, and daily volume, the phrase "price discovery" is a melody played on one instrument.
The mirror reflects only what its operator allows it to reflect.
Core IV: The Four-Hat Problem
Let me introduce an analytical framework that has shaped my recent work on the "Trust Layer" guidelines for institutional integration. The credibility of a financial product degrades as the number of structural roles played by the issuing institution increases. Gate occupies at least four distinct positions in this product.
First, it is the issuer. It structures the Mirror Note, defines the terms, selects the pricing band, and decides the distribution mechanics. Second, it is the underwriter. It charges a 5% fee to place the product with its own users, a fee that traditional underwriters earn for distribution risk that Gate is not actually assuming. Third, it is the market operator. It runs the only venue where the token can trade, with fee schedules set at its own discretion. Fourth, and most troubling, it is the settlement arbiter. It determines what happens upon IPO completion, or IPO failure, or any "other circumstances" that may arise.
In traditional securities regulation, these roles are separated by law precisely because their combination creates dangerous incentive structures. A market maker with the power to set settlement terms is, in effect, the referee and the goalkeeper simultaneously. On Gate, the separation exists only as internal policies that users cannot see.
I spent six months in 2022, after my startup funding evaporated, patching legacy multisig wallets and contributing dozens of fixes to the Gnosis Safe repository. That period taught me something about how trust actually operates in this industry. It lives in the boring places: in audited code, in reproducible build artifacts, in multisig thresholds that require an honest majority, in transparent failure modes. None of those elements exist in this product. There is no code to audit. There is no on-chain settlement to verify. There is no external auditor to confirm the asset backing. There is a database entry on Gate's servers, supported by a promise from an entity that is simultaneously writing the rules, operating the game, and keeping score.
Core V: The Regulatory Convergence
Let's address the question that every serious analyst eventually reaches: securities law. American courts assess investment contracts through the Howey framework. Money is invested: yes, 10,000 USDT minimum. There is a common enterprise: yes, subscriber capital is pooled into a pre-IPO exposure vehicle. There is an expectation of profit: explicitly marketed. And the profits come from the efforts of others: entirely. The value depends on Moonshot AI's team, its product execution, its navigation of China's AI regulatory environment, and its success in reaching a public listing. By conventional analysis, this instrument checks nearly every box.
The "Mirror Note" framing is designed to walk a tightrope between classification as a derivative and classification as an unregistered security. But a derivative that tracks the value of a private company's shares still qualifies as a security in most major jurisdictions if it is offered to retail investors without registration or exemption. The product's silence regarding eligible user geographies is itself informative. If the product were cleanly structured for non-U.S. users, the announcement could simply say so. The omission suggests a deliberate ambiguity about regulatory exposure.
There is also the compliance veneer of GUSD. GUSD is a regulated stablecoin issued by Gemini under New York Department of Financial Services supervision. Its inclusion in this product offers a signal of institutional polish that does not actually extend to the underlying token. Using GUSD as the subscription currency does not change the legal character of the Mirror Note. It is the equivalent of wearing a designer suit to a meeting with a banker while requesting a loan backed by nothing.
The question of Moonshot AI's own position deserves attention. The company appears to have no involvement in the token's creation. It has no obligation to honor the reference, no obligation to facilitate settlement, no obligation to recognize any claim from tokenholders. If Moonshot AI's prospective underwriters or existing investors decide the product's existence complicates their IPO preparation, they could pressure Gate to shut the program down. And then what? "Applicable rules."
Contrarian: The Real Product Is You
Here is the counterintuitive conclusion that crystallized for me as I worked through the numbers: even if this product succeeds commercially โ even if Moonshot AI IPOs above a $50 billion valuation and Gate settles with immaculate precision โ the KIMI token may still be a net negative for the ecosystem. Because its actual purpose may never have been to provide access to Moonshot AI at all.
In a sideways market, user attention wanders. Capital needs a story to stay anchored. Gate is building a narrative bridge between the AI boom and tokenized private markets, but more importantly, it is building a reason for idle stablecoin deposits to remain inside its walls. The 3.8% APR is not an act of generosity; it is a retention payment. The pre-IPO token is the story that makes the retention feel like an opportunity instead of a lock. For every subscriber who would have withdrawn their USDT in search of yield or opportunity, the product offers a reason to stay. The platform's own risk is not misaligned with yours โ it is orthogonal to yours. Gate earns fees whether the token rises or falls, whether the IPO happens or fails, whether you exit in profit or loss. It is earning fees from your capital while your capital waits, and it will earn fees again when your capital leaves.
We didn't build a future; we built a mirror. A mirror creates nothing. It only reflects, with distortion determined by the glass and the angle.
And the deeper lesson is this: the product asks you to measure the Moonshot AI story, but the discipline of financial engineering demands that you measure the instrument instead. The instrument is a centralized, opaque, high-fee, walled-market claim on a private company's uncertain future. Root: it is, in every meaningful sense, not the thing it reflects.
Takeaway
Open source is not a license; it's a state of mind. Nothing is further from that state than a Mirror Note that shows no receipts. The right pre-IPO instruments for this industry will eventually exist. They will carry audited custody, on-chain evidence of asset backing, transparent fee schedules, and exit routes into competitive markets. They will treat users as principals, not as sticky deposits. Until then, we are mining for truth in the noise of pre-IPO mania. And the truth is fairly simple: if you cannot verify the mirror, what you are buying is the mirror-maker's promise. In a market that survived 2022 by finally learning to ask hard questions, that promise is no longer enough.