Oracle has done what every old tech giant dreams of doing. In September, it announced a $300 billion cloud deal around OpenAI, the most recognizable name in software, and watched its stock soar.
Two months later, the market issued its verdict. Oracle has lost more than $300 billion in market capitalization, trading below pre-AI announcement levels, while reports have begun calling it the “ChatGPT curse.”
Analysts are now treating the massive deal as a case study of what happens when the promises of artificial intelligence exceed the cash flows it is supposed to support.
Meanwhile, Cursor just raised $2.3 billion at a valuation of $29.3 billion. The company surpassed $1 billion in annual revenue this year and beyond Tripled Evaluated since June.
The programming tool has unloaded investment capital on the promise that engineers will live inside an AI binary programmer that will write most of the code for them.
A development tools startup and a general software company are suddenly part of the same mental spreadsheet as most L1 tokens, and investors are now asking a slightly rude question.
When AI can hand a three-year-old startup a price of $29.3 billion, is the money still needed in crypto at all, or are cryptocurrencies being pulled into the same trade under a different bar?
Artificial intelligence money hose
A nice closer look at the crazy funding numbers explains this mood.
Funding startups in the field of artificial intelligence globally receipt About $100 billion in 2024, nearly 80% more than in 2023 and nearly a third of total venture capital that year. S&P Global places productive AI funding at more than 56 billion dollars In 2024, nearly double the previous year.
The Stanford AI Index tracks private investment in generative artificial intelligence in… $33.9 billion For 2024, more than eight times for 2022. EY Estimates In just the first half of 2025, AI startups raised another $49.2 billion.
Crypto remembers what that looks like. In 2021, the hot trades were token issuance, DeFi yield, and Metaverse stocks. In 2024 and 2025, the center of gravity has moved. Large checks were performed on training operations, data centers, and a small circle of foundational modeling laboratories. Barron’s counts nearly a third of global venture capital going into AI names like xAI, Databricks, Anthropic and OpenAI.
On the public side, companies are racking up huge debt piles to chase GPU capacity. Oracle is reportedly raising about $38 billion in bonds to fund the buildout of its cloud. Nvidia’s data center revenues have reshaped entire stock indices. If you want exposure to “future cash flows from computing,” the highest beta is now in the core models and infrastructure of AI.
This does not mean that liquidity from cryptocurrencies disappears. This means that marginal dollars are priced against a new benchmark. If a mid-sized AI startup commands a $30 billion valuation, and OpenAI can talk about trillion-dollar capital plans without being ridiculed, the barrier for a $10 billion code with little real-world use gets even higher.
Artificial intelligence codes and ASI experience
Crypto did the logical thing: it tried to package artificial intelligence inside tokens. The main effort was the AI Superintelligence Alliance, a plan to merge SingularityNET, Fetch.ai, and Ocean Protocol into a single ASI token and describe the entire group as decentralized AI. Fetch.ai Integration blog He put out a simple sales pitch in 2024. One vault, one token, and three projects that claim to cover agents, data, and models.
This worked for a while. Billions of dollars worth of AGIX, FET and OCEAN liquidity were channeled into the same narrative. Exchanges have arranged spot and perpetual pairs for ASI. Hash holders received migration bridges and one token that was clearly assigned to “AI” in the watchlist. Cryptocurrencies seem to have found a way to compress a chaotic sector into something that can live on a single line of a derivatives blotter.
Then the ocean walked.
In October, the Ocean Protocol Foundation announced its withdrawal from the alliance, requesting that OCEAN be separated from ASI and re-listed as a separate asset.
circumference framed Going out as a “voluntary union”. Fetch.ai has since launched legal action, with court filings tracking transfers of more than 660 million OCEAN to FET and alleging broken promises related to the merger.
This little governance drama tells you something about AI token trading. It’s chasing the same story as the private AI boom, but with more volatility and essentially no revenue. When ASI was trading well, everyone wanted to join. When evaluations subsided and community politics resurfaced, the “coalition” reverted to being three tables with different agendas.
From a liquidity standpoint, AI tokens look less like a separate asset class and more like a way for money in cryptocurrencies to shadow what is happening in their own AI. Latest Index Round or New Anthropy Round Finance From Amazon ASI doesn’t move on a strict basis, but they set the emotional tone. Cryptocurrency traders monitor stock trades and price their AI baskets accordingly.
From Bitcoin mines to AI model farms
The clearest merger between AI and cryptocurrencies lies in energy contracts. Bitcoin miners have spent a decade building data centers in cheap energy regions, and now AI miners are paying top dollar for the same megawatt base.
PetFarms It is the most obvious case. The company announced plans to completely scale back Bitcoin mining by 2027 and redeploy its infrastructure into artificial intelligence and high-performance computing.
Its 18-megawatt facility in Washington state will be the first site to be converted, with racks designed for Nvidia GB300-class servers and liquid cooling capable of handling about 190 kilowatts per rack.
Bitfarms’ press release describes a fully funded agreement worth $128 million with a large US data center partner. Management claims that a single AI facility could outperform the entire company’s historical Bitcoin mining profits.
Bitfarms is not alone. Iris Energy It was rebranded as IREN and it is Transformation Its hydro-powered sites into AI data centers, with Bernstein research pointing to billions in revenue expected from Microsoft-backed GPU deployments.
Hut 8 It speaks publicly about being the first powerhouse platform that can direct 1,530MW of planned capacity to whatever workload drives the best, with AI and HPC at the top of the list.
Basic scientific AI cloud provider CoreWeave has come far enough down this road Agreed A $9 billion deal to buy all the shares, aiming to secure more than a gigawatt of data center capacity for Nvidia’s heavyweight groups, before shareholders backed out.
The pattern is the same in each of these cases. Bitcoin mining has given these companies cheap power, grid connections, and sometimes hard-earned permits.
Then AI came along and offered a higher price per megawatt. For shareholders who have watched multiple halvings squeeze mining margins, directing power into GPU stacks clearly looks like trading a mature carry trade for growth.
This is where the headline “AI is eating crypto liquidity” becomes literal Bitcoin. Every megawatt that goes from SHA-256 to GB300 or H200 is a unit of energy that no longer secures the network. The hashrate has continued to grow as new miners come in and old ones retire, but over time, a higher share of the cheap energy will be priced out by the AI’s willingness to pay.
When AI attacks bars
There is another intersection between AI capital and cryptocurrencies: security.
In November, Anthropic published a report on what it described as the first large-scale espionage campaign organized by an artificial intelligence agent. A China-linked group jailbroken the company’s Claude Code product and used it to automate reconnaissance, exploit development, credential collection, and lateral movement across nearly 30 victim organizations.
Some attacks have been successful. Some failed because the form caused hallucinations of fake credentials and stole documents that were already publicly available. But the most disturbing part is that most of the attack chain was driven by natural language prompts rather than a room full of operators.
Cryptocurrency exchanges and custodians fall in the middle of this explosion radius. They already rely on AI within trading monitoring, customer support and fraud monitoring.
As more operations shift to automated agents, the same tools that route orders or monitor money laundering operations will become targets. The dense concentration of hot keys and wallets makes it attractive to any group that can direct a client of Cloud’s size to the network map.
The regulatory response to this type of event will not care whether the affected venue is trading Nvidia stock, Bitcoin, or both. If a major AI-based hack occurs at a major exchange, the policy conversation will treat AI and cryptocurrencies as a single risk surface that sits on top of critical financial infrastructure.
So is AI really consuming cryptocurrency liquidity?
The honest answer is that AI is doing something much more interesting. It sets the price of risk for anything that touches the account.
Investment money that might once have been chasing L1s is now funding basic AI models and infrastructure. Public equity investors are weighing a 30% divestment in Oracle against the possibility that the $300 billion OpenAI cloud deal could actually come to fruition.
Private markets are happy to rate a development tool like Cursor on par with a mid-sized token network. Bitcoin miners are rebranding as data center operators and signing long-term contracts with super miners. Token projects are trying to install “artificial intelligence” on their ticker because that is where the excitement lies.
Looking at this market from the depths of the cryptocurrency industry makes it look like a food chain where AI simply devours everything.
But unfortunately, it is always more subtle and complicated than it seems. Over the past couple of years, AI has become the reference trade for future computing, and this trade is dragging Bitcoin infrastructure, AI tokens, and even security exchanges into the same story.
So the liquidity does not leave immediately. It’s moving, pricing everything else against the one sector that convinced markets to fund trillion-dollar capex plans on the basis of promise and demonstration.
