On May 15, 2025, Ramp—a corporate expense management platform—released data claiming Anthropic leads U.S. enterprise AI adoption. This is not a tech review; it is a paid-adoption signal. For crypto allocators, this data point carries weight beyond the AI sector. It signals a shift in capital allocation from speculative token markets to productive AI infrastructure. The question: does this macro flow benefit crypto, or does it drain liquidity?
Ramp’s data is derived from enterprise software procurement records—bills paid for API access, SaaS subscriptions, and cloud marketplace purchases. The core fact: Anthropic’s Claude models are being adopted at a rate that surpasses OpenAI’s GPT suite among Ramp’s customer base. This is a concrete observation of corporate budget deployment. I have seen this pattern before. In 2017, I audited 400+ ERC-20 smart contracts during the ICO boom. The structural flaw then was that capital flowed into unverified protocols without standardised risk assessments. Today, the flaw is that capital is flowing into a single AI vendor without a comparable audit of its long-term viability. The macro implication is clear: marginal corporate dollars are rotating out of experimental crypto budgets and into proven AI services.
Context: Global Liquidity Map
To understand the significance, we must map the current liquidity environment. AI and crypto are competing for the same risk capital pool. In Q1 2025, global venture capital funding for AI hit $45 billion, while crypto VC struggled at $8 billion—a 5.6x gap. Stablecoin supply has been flat since March, hovering around $150 billion, indicating no net new liquidity entering the crypto ecosystem. The Federal Reserve’s rate trajectory remains uncertain, with the market pricing in a 50% chance of a cut by September. In this environment, any signal that enterprise budgets are prioritizing AI over blockchain further tilts the allocation equation.
Ramp’s data is not a random sample. It covers mid-to-large enterprises that use expense management software—typically tech, finance, and professional services firms. These are the same firms that once allocated 5-10% of their IT budget to crypto pilots (e.g., treasury diversification, DeFi yield, NFT collectibles). Now, that budget is being redirected to AI API costs. Based on my experience designing compliance frameworks for a Hong Kong-based fund in 2024, I observed a clear pattern: institutional clients were cutting crypto exposure in favor of AI infrastructure. The Ramp report quantifies what I saw anecdotally.
Core: Crypto as a Macro Asset Under AI Pressure
I treat the Ramp/Antitropic signal as a leading indicator for a liquidity rotation. The core insight: enterprise adoption of AI is creating a structural demand for compute, which in turn raises the opportunity cost for proof-of-work mining and GPU-based token issuance. On-chain data from April 2025 confirms this. Ethereum gas fees dropped 30% month-over-month, averaging 15 gwei. Meanwhile, AWS GPU instance prices rose 12% for the same period, driven by enterprise AI workloads. The correlation is inverse. As capital flows to AI, it leaves crypto.
But this is not a uniform effect. Specific crypto sectors may benefit from AI’s infrastructure demand. Decentralized physical infrastructure networks (DePIN) like Render Network and Akash Network could see increased usage as AI workloads seek cheaper, non-censored compute. However, the Ramp data shows that enterprises overwhelmingly prefer centralized, compliant AI providers—Anthropic, OpenAI, and Google. The market for decentralized compute is still nascent, with total locked value under $500 million across all DePIN projects. My 2020 DeFi liquidity stress-testing model taught me that when capital flows into one asset class, it exits another unless there is a direct bridge. The bridge between AI and crypto is weak today.
Let me break down the numbers. Anthropic’s enterprise adoption is likely concentrated in developer tooling and customer support automation. According to industry estimates, Claude API calls generate $0.03 per 1,000 tokens for output. If Ramp’s data reflects a 20% quarter-over-quarter growth in Anthropic spend, that implies a $150 million annualized revenue run rate from Ramp’s customer base alone. Extrapolate to the broader market, and Anthropic’s enterprise revenue could be $2-3 billion annually. Compare that to the entire crypto DeFi ecosystem, which generated $1.8 billion in fees in 2024. The AI enterprise market is already larger than the entire DeFi fee market, and growing faster.
This is not a prediction of crypto’s demise. It is a structural observation. The macro data demands a rebalancing of portfolio allocations. I have seen this movie before. In 2022, during the Terra-Luna collapse, I led a forensic audit of a $2 billion hack. The lesson was that liquidity cascades are unforgiving. When a large capital pool starts moving to a new asset class, the old class suffers a liquidity drain that amplifies volatility. The crypto market is now experiencing a slow-moving liquidity cascade to AI.
Contrarian: The Decoupling Thesis
The counter-intuitive angle is that enterprise AI adoption may decouple from crypto, but that decoupling is not uniformly negative. AI tokens, such as Render (RNDR) and Akash (AKT), could benefit from the narrative that AI needs decentralized compute. However, the Ramp data suggests otherwise. Enterprises prefer centralized, audited, and legally compliant providers. Anthropic has a clear security-first brand, which aligns with corporate risk management. Decentralized compute networks lack the compliance frameworks that large enterprises require. My 2024 experience designing institutional onboarding pipelines for a Hong Kong fund showed that the biggest barrier for crypto adoption is regulatory clarity, not technology. The same applies to AI: enterprises will pay a premium for a provider they can sue.
Furthermore, the decoupling thesis assumes that AI adoption will eventually spill over into crypto. I see the opposite: AI creates a separate, self-sustaining capital market that captures the marginal dollar. The crypto market’s recent sideways movement—with Bitcoin stuck between $60,000 and $70,000 for 90 days—is consistent with a liquidity drain. The Ramp data is another confirmation that the wave is not coming to crypto; it is going to AI.
Takeaway: Positioning for the Next Cycle
The forward-looking judgment is clear: the next six months will test whether crypto can maintain its own narrative or become a sideshow. I recommend positioning for divergence. Overweight assets that directly benefit from AI infrastructure demand—GPU tokenization, DePIN, and data storage protocols. Underweight speculative tokens that rely on retail liquidity and narrative cycles. The market is not pricing in the structural liquidity drain from AI.
We do not predict the wave; we engineer the hull. Volatility exposes weak balance sheets. Efficiency punishes sentiment. The Ramp report is a data point that demands a portfolio review. The macro environment is shifting. The question is not whether AI will dominate enterprise budgets—it already does. The question is whether crypto assets can find a parallel, non-competing value proposition. The answer, based on current on-chain data and capital flows, is not yet. Prepare for a prolonged period of capital rotation, and adjust your exposure accordingly. The next cycle will belong to those who read the macro signals, not those who chase the narrative.
