Looking back on June 20, 2026, it is hard to simply categorize the three signals that emerged during that brief period as mere "coincidences": according to a single source, after meeting with Anthropic CEO Dario Amodei during the G7, Trump stated that he no longer considers the company—which had once been viewed as a potential national security risk due to the Claude "jailbreak" controversy—to be a threat; almost within the same time frame, Nansen CEO Alex Svanevik publicly reminded that the current AI infrastructure, which relies on the scarcity premium of American computing power and models, may have its bubble burst if companies effectively adopt Chinese large models on a large scale; shortly thereafter, on-chain analyst il.hl presented a set of extreme numbers: as a decentralized derivatives trading platform, Hyperliquid reportedly recorded around $790 million in revenue and a valuation of approximately $20.6 billion with a team of about 14 employees, resulting in an average revenue of around $56.42 million per person and a profit margin close to 100%. Although still lacking official financial reports or third-party audit cross-verification, these numbers are significant enough to provide a snapshot of the efficiency of on-chain financial infrastructure. As the boundaries of AI security regulation are redrawn, the scarcity premium of AI infrastructure is questioned by insiders, and efficiency samples from on-chain derivatives platforms come to light, we must examine these three events together: are the seemingly parallel infrastructure tracks of AI and crypto heading toward a simultaneous recalibration of valuations?
Trump No Longer Targets Anthropic
Before the G7, Trump's opposition to Anthropic was rooted in a deep rift that arose from the "jailbreak" issue. The jailbreak vulnerability of Claude's models was considered by Trump as a potential national security risk, whereas Anthropic provided a much more moderate assessment in terms of risk rating, mitigation pathways, and public communication. For the United States, which has already raised concerns of national security to the level of export controls on large models and AI safety reviews, this divergence is no longer just a discussion between technical teams and red teams; it directly touches the national security red line of "which models, in whose hands, and in what forms they operate." Trump once included Anthropic in a list of potential threats, essentially responding to this uncertainty in the most conservative way possible.
A turning point occurred during the G7. According to a single source, after the summit, Trump met with Anthropic CEO Dario Amodei and stated that he no longer views Anthropic as a national security threat. Looking back on June 20, 2026, this statement, akin to “dropping charges,” effectively redraws the boundaries of U.S. AI security at a time when major global economies are engaged in a game around AI regulation: risk is no longer simply tied to a specific company or model but is more likely to be defined in terms of specific capabilities, uses, or flows. For regulators, this leaves greater flexibility within existing security review and export control frameworks, allowing for more frequent oscillation between “blanket prohibition” and “controlled openness” in evaluating large models; for companies, the released signal is that as long as they can be shaped into “controllable and negotiable partners” within regulatory discourse, there remains a strategic window to reshape their risk profiles through compliant design and policy lobbying within a high-pressure security framework.
Where the Bubble Begins to Burst
At the same time when the regulatory discourse is being reshaped, Alex Svanevik turned his focus to another, more fragile link: the valuation narrative itself. His judgment is not predicting a "last sell order," but rather reminding the market that the current high valuation of AI infrastructure is based on the premise that "American computing power and model supply are of irreplaceable scarcity." Once global companies begin to effectively adopt Chinese large models in real business scenarios, the American solution will no longer be the only viable option, thus rewriting this scarcity premium, and the bubble will begin to deflate from the level of valuation assumptions rather than being reflected in daily prices first.
In his narrative, Chinese models do not need to break through U.S. regulation to erode this premium. Even under tightening export controls and security reviews, as long as Chinese large models can rely on globally available GPU supply and non-cutting-edge hardware to operate stably in other compliant environments, providing acceptable performance and cost combinations, the narrative of American AI infrastructure being "irreplaceable" begins to distort. Once the capacity of computing racks and chip production continues to expand, "scarcity" itself becomes an asset that diminishes over time, and the real danger facing the entire industry is the moment when the market collectively realizes that this premise has been quietly rewritten.
$790 Million Generated by 14 People
If the "high valuation + heavy assets" story of AI infrastructure is distorting, then the on-chain world is presently telling another extreme narrative. According to on-chain analyst il.hl, Hyperliquid, as a decentralized derivatives trading platform, has an estimated annual revenue of approximately $790 million and a corresponding valuation of about $20.6 billion, based on publicly available information that only cites around 14 employees. This translates to an average revenue of nearly $56.42 million per person, and the same source also claims its profit margin is close to 100%. In the comparative samples, Hyperliquid's per capita output is reported to be significantly higher than that of the CME and most mainstream crypto trading platforms; this combination of “small teams + high revenue + high profit margin” is nearly unimaginable within the paradigms of traditional finance and most crypto institutions.
However, this "efficiency myth" currently relies on a single on-chain analyst's perspective: whether it's the $790 million in revenue, the $20.6 billion valuation, or the team of 14, and the nearly 100% profit margin, all lack official financial reports or third-party audit cross-verification, and can only be viewed as a strong signal that requires follow-up comparison rather than conclusive proof. It reveals the extreme possibilities of on-chain financial infrastructure regarding cost structure and marginal expansion yet also reminds participants that until more transparent disclosures are made, any valuation or narrative based on this data remains at the level of high uncertainty assumptions.
AI Valuation and On-Chain Platform Repricing
On the opposite side of this set of "extreme efficiency" data is a completely different valuation story for the leading AI infrastructure: its current market value is more dependent on "computing power scarcity + regulatory moat," with the market willing to pay in advance for future computing rents and service fees protected by regulatory frameworks. Alex Svanevik's warning happens to hit here—once enterprises begin to effectively adopt Chinese large models, the scarcity premium of American computing power and models will be weakened, and today's high valuations surrounding American infrastructure could shift from "irreplaceable" to "replaceable," as valuation anchors slide from policies and regulations to more conventional cost, performance, and price comparisons.
In contrast, for on-chain financial infrastructures like Hyperliquid, if the circumstances reported by il.hl hold true, then its $20.6 billion valuation is built on approximately $790 million in revenue, a very small number of employees, and a profit margin close to 100%. The valuation discussion initially focuses on the current cash generation capability of "high per capita and high profit margin," rather than following the traditional route of heavy asset expansion or long-term losses in exchange for market share. This "code + protocol" driven high-efficiency model appears closer to a sustainable business paradigm for infrastructure: low marginal costs, a minimal organization, and relatively little hard dependency on a single jurisdiction. However, it must be emphasized that the current disclosures regarding Hyperliquid's revenue and profit margin still rely on a single source, lacking official financial statements and third-party audit cross-verification, and standing at a crossroads of policies, technology, and enterprise adoption uncertainty in both AI and on-chain tracks, any new valuation formulas based on these numbers resemble an ongoing experiment rather than a concluded certainty.
Next Steps from Policy and Valuation Perspectives
Looking back after June 20, 2026, Trump's statement post-G7 about no longer viewing Anthropic as a national security threat (from a single source) is hard not to interpret as a subtle shift in the U.S. AI regulatory narrative: in the future, it is more likely to shift from "one-size-fits-all security threats" to a combination of "graded risk control + export controls," easing security labels for some companies while continuing to lock the transnational flow of computing power and top models with national security logic. Parallel to this, Alex Svanevik clearly points to the trigger for the "AI bubble" as "Chinese models being effectively adopted by enterprises," with the true determinant of whether the bubble bursts being not just one or two valuation adjustments, but to what extent Chinese models can traverse regulatory gaps and form alternative enterprise-level supply supported by global GPU and non-cutting-edge hardware availability. This path is naturally filled with time lags and policy uncertainties. For on-chain platforms represented by Hyperliquid, what is more worth monitoring next is not the unverified $790 million revenue and nearly 100% profit margin data from a single source, but four more sustainable observational dimensions: whether the team expands rapidly under high valuation pressure, whether fee and subsidy structures are forced to be rewritten, whether on-chain activity and protocol usage can be maintained after the narrative retracts, and how their compliance posture in different jurisdictions adjusts as regulations tighten. AI and on-chain financial infrastructure are currently locked at the intersection of "regulatory tightening + global competition," and which entities can preserve their valuations ultimately depends on their dual responses to real enterprise adoption and sustainable income in both on-chain and offline, and the quality of this response represents the largest open variable for the next round of valuation repricing.
Join our community to discuss and grow stronger together!
Exclusive Hyperliquid benefits for AiCoin: https://app.hyperliquid.xyz/join/AICOIN88
Exclusive Aster benefits for AiCoin: https://www.asterdex.com/zh-CN/referral/9C50e2
On-chain Telegram community: https://t.me/AiCoinWhaleData
On-chain community: https://www.aicoin.com/link/chat?cid=N6OVMor5g
AiCoin on-chain Twitter: https://x.com/aicoinwhaledata
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
