Elon Musk wants to add 10GW of computing power by the end of 2027. That's enough to run the entire Bitcoin network 10,000 times over, or to train a frontier AI model every 36 hours. But as a crypto educator who's spent years dissecting the failures of centralized systems, I see a deeper problem: we are about to hand the keys to the most powerful infrastructure ever built to a single company. And the crypto community is cheering it on.
The Numbers That Matter
A SemiAnalysis report released last week revealed the scale of SpaceX's ambition. Musk's conservative target is 6-8GW of incremental computing power in 2027, with upside exceeding 10GW. At a capital expenditure of roughly $50 billion per gigawatt, that means $300-500 billion in spending next year alone. For context, the entire global cloud computing market spent about $200 billion in 2025. SpaceX's buildout, if realized, would dwarf the current hyperscale providers.
But the revenue projections are even more staggering. SemiAnalysis models that when OpenAI and Anthropic provide API inference services on Nvidia GB300 clusters, each GW can generate over $100 billion per year—at a rental cost of only $3 per GPU per hour, or about $12 billion annually per GW. That's a margin that would make any DeFi yield farmer blush. Microsoft's $250 billion infrastructure deal with OpenAI from October 2025 corresponds to roughly 7GW. SemiAnalysis estimates SpaceX could sign a ~3GW contract with Microsoft worth $150 billion, pushing SpaceX's annual recurring revenue to $300 billion by end of 2027.
These are not normal numbers. They are the kind of hockey-stick projections that usually accompany a crypto whitepaper pitch. But unlike most crypto projects, SpaceX has a track record of delivering on audacious hardware goals. The question is: what does this mean for the decentralized web?
The Centralization Trap
Let me be clear: I am not anti-compute. I've spent years teaching people about blockchain, and the single most important lesson is that compute is not the scarce resource—trust is. Truth is not mined; it is remembered. But when a single entity controls 10GW of the world's most advanced silicon, the balance of power shifts irrevocably.
Consider the current state of Bitcoin mining. After the fourth halving, miner revenue collapsed, and hash power has concentrated into three pools. The network remains secure, but the ideological promise of 'one CPU, one vote' is dead. Now imagine that same dynamic applied to AI inference. If SpaceX becomes the default provider for the most valuable AI workloads, they hold the keys to the most powerful non-human intelligence we've ever created. They can decide which models run, which data is processed, and—most dangerously—which governance rules apply.
In the chaos of the chain, find the signal. The signal here is that compute centralization is the new regulatory capture. Just as we fought against Wall Street gatekeepers, we must now question whether a single company should own the infrastructure that powers the next generation of autonomous agents, decentralized science, and—yes—crypto's own scaling solutions.
The Crypto Blind Spot
We in the crypto community have a tendency to celebrate any form of infrastructure growth. More compute means faster transactions, cheaper L2 rollups, and better zk-proofs. But we forget that the ethos of decentralization is not just about ownership—it's about distribution of power. Culture is the new consensus mechanism. If SpaceX's computing power is used to train AI models that then replace human decision-making in DeFi, we've simply swapped one central authority for another.
I've seen this pattern before. In 2022, during the Terra collapse, I ran a series of post-mortems on failed protocols. The common thread was not bad code, but bad philosophy. Projects that promised 'code is law' but built in backdoors for the founders. SpaceX's compute buildout is not a backdoor—it's a front door. They are building a fortress of silicon, and they are the only ones with the keys.
The Contrarian Angle: Compute as Commodity
Now, let me play devil's advocate. Perhaps this massive compute buildout is exactly what crypto needs. The SemiAnalysis report assumes that SpaceX will sell compute at market rates. If they do, it could drive down the cost of GPU time, making decentralized AI projects like Render Network or Akash more competitive. More compute could also accelerate the development of fully homomorphic encryption and other privacy-preserving technologies that crypto desperately needs.
But here's the catch: SpaceX is not a public utility. They are a private company with a history of vertical integration. They own the rockets that launch the satellites, the satellites that beam the data, and—soon—the data centers that run the compute. We do not build walls; we build bridges for value. But if SpaceX builds a walled garden of compute, they will capture the value, not the community.
Moreover, the revenue projections are based on a fragile assumption: that AI inference demand will remain concentrated in a few large players like OpenAI and Anthropic. If the market shifts toward smaller, decentralized models running on edge devices, the demand for hyperscale clusters could evaporate. Remember the ICO boom? Everyone thought they needed their own blockchain. Then L2s and rollups proved that shared security is more efficient. The same logic applies to compute: maybe we don't need 10GW in one place; we need 10GW distributed across a thousand nodes.
The Takeaway
The SemiAnalysis report is not just a financial analysis—it's a warning. The future of computing is being decided right now, and the crypto community must engage with the reality of infrastructure centralization. We cannot claim to build a decentralized world while cheering on the most concentrated compute buildout in history.
The future is written in code, but felt in spirit. If we want a spirit of decentralization, we must demand that compute power is distributed, open, and accessible. Not locked inside a single company's data center. The bridge we need to build is not between SpaceX and Microsoft—it's between the promise of decentralized technology and the reality of who controls the machines.
As I tell my students: 'Ideas have no gas fees, only gravity.' The idea of a decentralized internet is beautiful, but it has no weight unless we build the infrastructure to support it. SpaceX's 10GW is a test of whether we will build that infrastructure together, or hand it over to a single builder.