On the evening of July 16, AI large model company Moonshot AI officially launched its next-generation flagship model, Kimi K3.
2.8 trillion parameters, the world's first open-source 3T-level model, topping frontend code capability tests and outperforming Claude Opus 4.8.
According to media reports from 21st Century Business Herald, Securities Times, and others, following the release of K3 and amid renewed market repricing of AI capital expenditures, the U.S. AI industry chain collectively lost approximately $470 billion in market value. The Philadelphia Semiconductor Index plunged 12.5% in one week, officially entering a technical bear market. NVIDIA dropped 5.3% in a single day, while the Nasdaq Composite fell 1.5%. Across the Pacific, according to financial media BusinessFocus, Zhipu, hailed as the "first listed company in large models" on the Hong Kong Stock Exchange, plummeted 28.49% on the day of K3’s release, with a cumulative decline of over 40% over two days; MiniMax also suffered a sharp drop of more than 15%.
Yesterday, according to U.S. tech media Axios, citing informed sources, the Trump administration is considering targeting Chinese AI models—pursuing legal liability against U.S. cloud providers that host or operate open-source Chinese models, alongside threats of export control blacklists. It should be noted that this remains in the "considering" stage and has not yet become formal policy or an announced measure. During the same trading session in which the news broke, on-chain markets saw nearly simultaneous significant declines.
Pricing power, who moves first? Beyond traditional markets, there’s another arena: Hyperliquid, an on-chain perpetuals platform where traders can speculate on the valuation of both public and private companies. However, it’s important to clarify that these contracts on Hyperliquid trade expectations of future company valuations—they are perpetual derivatives, not equity, not funding prices, and not official valuations. They reflect the market’s anticipated price level, not the company’s intrinsic value.
During the same trading session in which the White House announcement was made, Zhipu on Hyperliquid dropped 16.35%, Anthropic fell 12.76%, MiniMax declined 11.68%, and OpenAI slid 6.80%—the contracts for all four assets nearly plunged in unison. The on-chain market showed almost no delay in reacting, as sentiment toward these companies was immediately adjusted.
The most intriguing aspect of these numbers isn’t the decline itself, but the "identities" of these four companies. Zhipu and MiniMax are publicly listed companies with tradable stock prices on the Hong Kong Stock Exchange; Anthropic and OpenAI remain private, with their private-market valuations still undetermined. Two Chinese companies, two American companies—listed and unlisted, Chinese and American—the market’s valuation expectations for all four have been synchronously adjusted on-chain in unison.
This is not a coincidence; it is a structural inevitability.
Pricing in traditional markets follows an entire process: news is released, analysts work overnight to write reports, exchanges open, market makers place orders, and capital shifts across time zones. The sharp decline of Zhipu and MiniMax in the Hong Kong market was gradually realized only after Friday; the wipeout of the U.S. semiconductor sector was also carried out in stages over the course of a week.
There is no such process on-chain. There are no opening hours, no price limits, no time zone differences across markets. A benchmark score or a retweet from an industry leader can trigger drastic fluctuations in funding rates within minutes—this is why valuation expectations for companies like Anthropic and OpenAI, which have never gone public and whose private market valuations were traditionally confined to a select few institutions, are simultaneously adjusted on-chain alongside listed companies like Zhipu and MiniMax.
Valuation expectations in the primary market are increasingly being influenced by on-chain prices. The boundary, once maintained by information asymmetry and access barriers, is becoming blurred.
The right to price discovery is shifting. The foundation of traditional finance's pricing power is an entire approval system: securities must be registered, listings must be reviewed, and cross-border transactions require compliant channels. This system has operated for over a century—efficiency is not its goal; stability and control are.
But blockchain infrastructure, combined with financial innovations built on top of it, plays an entirely different game—efficiency is the only KPI.
As soon as the information appears, on-chain markets complete price discovery within minutes; traditional markets must wait a full trading day—or even an entire earnings season—to absorb it. This generational difference in reaction speed is fundamentally a shift in who holds the power of price discovery—not who ultimately decides the value of these companies, but who can first turn market sentiment into a tradable number.
The first to set the price takes the lead in defining the narrative’s starting point. The reports written by Wall Street analysts the next day are, in a sense, merely validating the expectations already demonstrated through real-money trades by on-chain traders the night before.
This is what makes the Kimi K3 incident noteworthy—it’s not just a technological breakthrough in AI models; in an increasing number of similar cases, on-chain markets have become one of the first places to achieve expected pricing.
The sharp declines in semiconductor stocks and the setbacks for Zhipu and MiniMax were gradually priced in. This time, the on-chain market re-priced expectations for these companies faster than traditional capital markets.
When Regulatory Boundaries Meet On-Chain Markets Let’s start with a control case. Between the release of Kimi K3 and reports that the White House was considering targeting Chinese AI models, only a few days passed—any entity tied to dollar settlement, U.S. cloud infrastructure, or U.S. investors is theoretically subject to long-arm jurisdiction; the response was swift, and the tools at hand are plentiful.
But the same assets yield completely different results in a different market.
Zhipu and MiniMax are domestic companies and listed entities under China’s regulatory framework; the synchronized adjustment of valuation expectations on-chain, alongside Anthropic and OpenAI, also reflects the attention and capital participation of domestic investors in these assets. However, this pricing expectation was executed on an offshore, decentralized platform requiring no access approvals, beyond the reach of China’s regulatory framework.
Domestic regulations have long established clear guidelines for cryptocurrency trading, but these guidelines apply only to exchanges that are licensed, have physical operations, and can be summoned for dialogue. When it comes to a perpetual contracts market where betting and matching occur on overseas servers, anonymous wallets are used, and no counterparty identity verification is required, existing regulatory tools clearly reach their limits—not because there is no legal basis, but because cross-border enforcement is inherently difficult: jurisdiction, evidence collection, and identity verification all require crossing international borders.
Whether this regulatory framework can cover such offshore on-chain markets, and to what extent, still leaves considerable room for discussion. As a result, domestic large model companies are subject to existing regulatory frameworks and corresponding disclosure obligations within China, while market expectations for their valuations have already begun to shift on offshore on-chain markets.
This is not entirely a matter of capability, but rather a matter of differing regulatory frameworks responding at varying paces to new financial infrastructure. The U.S. can advance policies around areas where it has jurisdictional leverage, such as cloud services and chip supply chains; however, how existing regulatory frameworks can adapt to new price discovery mechanisms in offshore on-chain derivatives markets remains a significant issue worthy of attention.
Price moves faster than news Returning to the Kimi K3 itself.
It brings not only a leap in model capability but also a shift in business model. Moonshot has priced K3 close to the level of Anthropic’s Sonnet series, signaling that it is no longer targeting price-sensitive retail users, but rather high-end clients such as banks, governments, and multinational corporations—directly competing with OpenAI and Anthropic.
Meanwhile, market rumors suggest that Moonshot is pushing for a Hong Kong stock exchange IPO, with its latest valuation reaching $30 billion, equivalent to approximately 100 times ARR. This multiple exceeds that of most high-growth software companies.
Investors are not betting on subscription revenue, but on whether Moonshot can grow into China’s version of Anthropic and secure long-term pricing power in the enterprise API and open model markets.
On-chain, forward-thinking capital has already voted with its feet—moments after the Axios report broke, the contracts for Zhipu, Anthropic, MiniMax, and OpenAI all dropped in sync, far outpacing traditional financial media’s ability to analyze and publish market impact reports.
The time lag in between is the entire secret behind this wealth transfer. Whoever can interpret the on-chain price signals gets the script before traditional media and analysts do.
Leave a question The "DeepSeek Moment 2.0" sparked by Kimi K3 may ultimately turn out to be a false alarm, just as it did last year—major cloud providers continuing to increase capital expenditures, and concerns over chip demand once again proving unfounded.
But one thing is now irreversible: the valuation boundary between the primary and secondary markets, once maintained by information asymmetry, is being rapidly eroded by on-chain expectation pricing tools.
Valuation speculation in private companies was once a game for a select few; now, anyone who opens Hyperliquid can place bets on the future valuation of a private company.
Will the next real market consensus form first on exchanges or on-chain?
This question is for those still waiting for the opening bell.
The content in this article is for reference only and does not constitute any investment advice. The market carries risks; investments should be made with caution.