Hook
A Wall Street analyst just valued SpaceX at $210 billion, not for its rockets or its Mars ambitions, but for its AI platform. In a recent note, Morgan Stanley's Adam Jonas argued that the company's "computing, connectivity, and real-time data capabilities" create an AI platform worth billions. The market cheered. But as someone who has spent years auditing the trust architectures of decentralized networks, I see a different story. The most valuable asset in this narrative—Starlink's global data pipeline—is being built as a centralized fortress. And that's exactly where the fragility lies.
Context
Let's unpack the report. Jonas separates SpaceX's valuation into three pillars: launch services, Starlink broadband, and an AI platform. The AI platform thesis rests on Starlink's ability to generate and process massive amounts of real-time data from its constellation of over 6,000 satellites, connected by laser links. The idea is that this data—combined with the compute power of AI models—can unlock new revenue streams in autonomous vehicles, defense, and global communications. However, the report conflates two distinct entities: Grok, the LLM developed by xAI, and SpaceX's own AI infrastructure. Grok is not built by SpaceX; it's trained on xAI's supercomputers. What SpaceX actually has is a sophisticated engineering AI for rocket landings and collision avoidance, plus a massive data transmission network. The real technological rarity is the laser inter-satellite links and the ground station mesh, not any model parameter.
Core
From a blockchain perspective, the Morgan Stanley thesis is a wake-up call about the value of data monopolies in the AI era. But it's also a cautionary tale. Starlink's architecture is a classic example of centralized control: one company owns the hardware, the data pipeline, and the AI layer. They can throttle, surveil, or monetize the data as they see fit. The network is not permissionless; the user terminals are locked to the network, and the routing is proprietary. This is the opposite of the decentralized ethos that underpins Web3. In my experience working with open-source protocols, trust is not something you can own—it's something you must earn and distribute. I recall a project in 2022 where I helped a DAO design an on-chain reputation system for data contributions. The core insight was that data provenance and access rights must be transparent and verifiable to build long-term trust. SpaceX's AI platform, by contrast, is a black box. The data flows through Starlink's satellites, but who owns the pipeline? Who ensures that the data is not used for surveillance or military advantage without community consent? The answer is: no one outside SpaceX.
The real value of the AI platform is not the compute or the connectivity—it's the data monopoly. With 6,000 satellites and 3 million subscribers, Starlink is generating a firehose of real-time geolocation, network usage, and environmental data. This is the fuel for any AI model that needs to understand the physical world. But in a centralized system, the data is a single point of failure. A government demanding access, a security breach, or a corporate decision to change the pricing model could all disrupt the entire ecosystem. We've seen this before in the crypto world: centralized stablecoins like USDC can freeze addresses within 24 hours. Circle's compliance-first approach is its biggest risk because it relies on a single authority to make decisions. Similarly, SpaceX's AI platform could become a tool for censorship or control, not liberation.
Contrarian
Now, the contrarian angle: maybe centralization is exactly what makes this valuable. In the short term, a vertically integrated AI platform can deliver reliability and speed that decentralized alternatives cannot match. Starlink's laser links offer low latency and high bandwidth, perfect for real-time AI inference in autonomous vehicles or drone swarms. A decentralized network of satellite nodes running on blockchain would be slower, more expensive, and harder to coordinate. The market is rewarding efficiency, not idealism. But here's the blind spot: the same centralization that creates efficiency also creates a single point of regulation. In the current bull market, every fond remembers the Terra collapse and the FTX debacle. We saw how quickly trust evaporates when a centralized entity fails. The AI platform valuation presupposes that SpaceX will remain benevolent and competent indefinitely. That's a dangerous assumption. Moreover, the military and defense use cases that Jonas highlights are precisely the ones that require the highest levels of trust and auditability. A decentralized, open-source space data network would be more resilient to cyberattacks and geopolitical pressure. As I write this, I think about the 2017 ICO boom in Hangzhou, where I organized literacy circles to help people see through the hype. The same lesson applies: when the value proposition is built on data, the architecture of that data's governance determines its long-term sustainability.
Takeaway
So where does this leave us? The Morgan Stanley report is a reminder that the intersection of AI and space is the next frontier for value creation. But the winners will not be the companies that own the most data—they will be the networks that allow users to own and control their own data. The next generation of space infrastructure must be built on decentralized principles, with transparent governance, permissionless access, and on-chain provenance. We need to start designing the protocols for satellite data sharing, not just the hardware. The question is not whether SpaceX can build an AI platform—it's whether we can build a trust layer that makes that platform accountable to its users. As I often say, bridges aren't built by a single hand—they're compiled, verified, and shared. The same is true for the data bridges that connect Earth to space. Let's build them together.
Code is only as strong as the trust it protects. Trust isn't created by a single entity—it's compiled, verified, and shared. Bridges aren't built by a single hand—they're compiled, verified, and shared.