On July 4, 2026, two seemingly unrelated news stories made headlines on the same day: On one side, Tether CEO Paolo Ardoino typed out several cold-hearted questions on X, directly pointing out that the AI industry is using subsidized computing power to achieve growth—giants rely on the mad expansion of GPU and server clusters to "feed" users while compressing the return cycle into an unclear future; he reminded that those high-priced computing assets depreciate economically in just 3 to 5 years, while Microsoft, Google, and Meta have already invested hundreds of billions in this area in recent years. If the profit model continues to remain ambiguous, the mismatch between capital return and depreciation cycles will eventually be accounted for by the market. On the other side, the police in the Val-de-Marne department of France held a press conference announcing the successful cracking of a cryptocurrency scam case involving about 1.5 million euros: a mother and son used a traditional method known as "Rip Deal" to fabricate transactions and concoct high returns, luring victims into transferring so-called "guarantee fees," and this time, the payment method was changed to cryptocurrency. The police tracked this case for nearly a year before peeling back the layers; globally, similar cases are increasingly tying old tricks to crypto payments, making tracing and evidence collection increasingly difficult, and exposing the reality of information asymmetry and weak user security education in cryptocurrency asset usage scenarios. The unsolvable accounts of AI computing power subsidies and the Rip Deal under crypto payments were presented on the same day, like two mirrors reflecting the same theme: beneath the technological boom, trust is quietly being overdrawn, while risk pricing has yet to keep pace with the acceleration beneath our feet.
Tether CEO Sounds the Alarm
As the Rip Deal news was still ripening, Paolo Ardoino's long post on X pointed the finger at another end—the "subsidy game." He named large tech companies for acquiring users through subsidized computing power—stacking massive GPU and server farms to exchange extremely low or nearly free AI services for scale and user stickiness. The issue lies not in the expansion itself, but in the “completely reliant on high capital expenditure-driven infrastructure frenzy” behind this expansion, while profit models and cash flows are still far behind chasing after it.
Ardoino specifically highlighted a hard constraint often overlooked in the clamorous narrative: the economic depreciation cycle of computing power-related assets is only about 3 to 5 years. In other words, the hundreds of billions invested by Microsoft, Google, and Meta in computing power over the past few years must find sufficient revenue to offset it within just a few reporting cycles; otherwise, these rapidly depreciating iron and silicon will ultimately become a burden on users and shareholders. Amidst the debate over whether there is a bubble in AI, this CEO, who has long been at the forefront of the crypto industry, chose to approach it from the mismatch between asset depreciation and capital returns, sending a colder reminder: when high CAPEX, high depreciation, and an unclear profit path are bundled together in motion, the so-called technological revolution may very well only be quietly shifting systemic risks forward. This reminder from within the industry acts more like a calm bell, urging all participants immersed in the AI feast to reassess: who will truly foot the bill for today's subsidized computing power tomorrow.
How Far Can the Myth of Growth from Computing Power Stacking Go?
"First grab the users, then think about how to make money," packaged as the commercial story of the "AI native growth flywheel," is not complicated in essence: on one end are companies like Microsoft, Google, and Meta investing hundreds of billions in data centers, GPU clusters, and network infrastructure, minimizing inference and training costs, leading the end users to see “free writing assistants” and “almost zero-threshold intelligent customer service”; on the other end, there are high capital expenditures quietly noted in the background, hoping to recoup this expense through enterprise-level services, subscriptions, and advertising over the years. The problem is that these computing power assets often face significant depreciation in accounting and economics in just 3 to 5 years, while many AI services are currently free or highly subsidized, resulting in extremely limited direct cash flow. When the depreciation curve and the profit curve misalign, the seemingly continuous rise in the number of users and calls is actually using time to exchange for an increasingly steep financial gradient.
This scene is not unfamiliar. The history of the internet and the cryptocurrency industry has seen the path of "first massively laying out infrastructure, then searching for profit models": laying down fiber optics first, getting nodes up and running, and then hoping that traffic and applications will naturally grow one day. The result is that multiple rounds of bubbles have burst to remind the market that without sufficient solid paid scenarios, infrastructure will ultimately become high-depreciation sunk costs. Today's AI expansion is replicating the same path—capital markets are betting valuations on future corporate contracts, subscription renewals, and advertising monetization, but before these revenue streams are truly operational, the more aggressive the computing power investment and the shorter the depreciation cycle, the faster the so-called "high-speed growth" is likely to be nothing more than a magnified mirror of subsidies. The growth myth created by the stacking of computing power can go as far as there are enough users willing to pay for it before the asset enters the end of its depreciation.
Breaking the 1.5 Million Euro Rip Deal Scam in France
On the same day that Paolo Ardoino questioned the computing power subsidy model on X, the Val-de-Marne police in France announced that they had cracked a "Rip Deal" cryptocurrency scam. The so-called Rip Deal is an old trick: scammers throw out a “high-value transaction” far exceeding market prices as bait, creating an illusion of "once-in-a-lifetime opportunity, massive profits ahead." Victims are invited to negotiate, view contracts, and conduct due diligence—all appearing as a normal large business transaction until the final "necessary step" is proposed—first to pay a guarantee fee to “lock in the transaction” or "prove financial capability." In this case, according to the Nice-Matin, the mother and son involved completely encrypted this step, demanding victims pay the so-called guarantee fee in cryptocurrency, accumulating around 1.5 million euros.
In traditional Rip Deals, cash and bank transfers might leave visible traces, but when the payment step is switched on-chain, the mother and son were able to quickly transfer chips and confuse the flow of funds after completing the induction. The Val-de-Marne police began the case around 2025 and only announced the resolution in 2026 after a whole year of unraveling the seemingly transparent yet highly anonymized on-chain fund path. The official description of "solved after a year of investigation" itself signals that when old scams meet new payment tools, the cost of law enforcement in terms of evidence collection and tracking is magnified many times, while ordinary people ignited by high returns at the beginning of the story often realize in the news of the case being cracked that their trust has long been overdrawn.
When AI Capital Mismatches Encounter Crypto Security Black Holes
If we put the victims in the French Rip Deal case, who were pushed along by layers of verbal tricks, on the same timeline as the users happily scrolling through "free" AI services in front of screens, we would find a strange symmetry: on one side is the bait of "high returns, shortcut transactions," leading people toward wallet addresses and private keys they cannot understand; on the other side are the giants using hundreds of billions of CAPEX to subsidize computing power, allowing a model so complex it is hard to audit, to seep quietly into every workflow in nearly zero prices or even for free. Paolo Ardoino stated on X that the current AI model is forming a "multiple structural mismatch"—with GPU and server depreciation cycles of only 3 to 5 years being used to gamble on an unproven profit path. This mismatch will not disappear but will only be transmitted at pressure points.
In the crypto world, the transmission is extremely direct: when the Rip Deal ties traditional face-to-face transactions and crypto payments together, the 1.5 million euros are nearly evaporated in an instant on-chain, and users lacking security education may not even understand what they authorized when clicking confirmation; in the AI industry, the transmission is more concealed—today's low prices and subsidies might be exchanged for tomorrow's sudden price hikes, reduced service quality, or more aggressive extraction and reuse of individual and corporate data. Whether it's AI giants telling the story of computing power expansion through "technological revolution" narratives or crypto criminals packaging scams under the banner of "new opportunities on-chain," both are leveraging the opacity of technology and the lag in regulation and law enforcement regarding high-complexity systems to obscure the real risk structure, outsourcing systemic vulnerabilities to terminal users who cannot see or negotiate their way out.
Who Will Restore Trust Before the Bubble Burst?
On July 4, 2026, when Paolo Ardoino questioned the risks of the AI industry piling up subsidies with high CAPEX while ignoring the mismatch between 3 to 5 years of depreciation and profit cycles, and on the same day, the French police announced the crackdown on a scam case that used "Rip Deal" rhetoric to lure victims into transferring about 1.5 million euros in crypto assets, the two seemingly unrelated news pieces stitched together a single proposition: technological narratives can inflate valuations and trading enthusiasm but cannot replace transparent risk pricing and effective protection for individual users. In the past few years, the AI computing power infrastructure has continuously drawn investments of hundreds of billions, while crypto payments are increasingly being grafted onto traditional fraudulent techniques globally, leading to long-term underestimation of risk and overconsumption of trust, which are common preambles to historical bubble bursts. For investors and ordinary users, facing any "AI+" or "on-chain+" opportunities, they must simultaneously pay attention to two lines: one is how capital recoups and in how much time, and whether the business model can close the loop within the 3 to 5-year economic lifespan of the assets; the other is where their asset security boundary lies, the extent of information asymmetry, and who will bear the risk if something goes wrong. What will truly determine the direction of this cycle is whether regulation can compel the AI and crypto industries to clarify risks and disclose real incentives and responsibilities without stifling innovation to the extent of a blunt knife, allowing technological applications to be reintegrated into a predictable and accountable trust framework rather than continue to overdraw the confidence of the next round of investors in the price of the narrative.
Join our community, let’s discuss together and become stronger!
AiCoin exclusive Hyperliquid benefits: https://app.hyperliquid.xyz/join/AICOIN88
AiCoin exclusive Aster benefits: 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。
