When Capital Becomes the Algorithm
On an ordinary Tuesday morning in Washington DC, I sat reading the wire reports with a familiar unease. OpenAI had just secured $122 billion in new funding. Not a Series F. Not a mega-round. A financial event the size of a small nation's GDP. Sam Altman, the company's chief executive, said something that should give every technologist pause: "AI compute is the most expensive project."
I have spent the last decade watching decentralized systems challenge centralized power. This was the opposite. This was centralization at planetary scale. And yet, the deeper story is not about money. It is about what money buys when the asset class is intelligence itself.
Bulls react. Bears reflect. The rest of us need to understand what just happened.
The Cost of Thinking at Scale
To understand why $122 billion matters, we have to understand what compute actually costs. When I audited blockchain whitepapers in 2017, the infrastructure conversation was about nodes and validators. A few thousand dollars in server costs. Today, the frontier AI models require training clusters that consume electricity like small cities. GPT-4 is estimated to have cost over $100 million in training alone. GPT-5, or whatever comes next, will be in a completely different order of magnitude.
Altman's statement about compute being "the most expensive thing" is not rhetorical. It is a structural truth. The AI industry has crossed a threshold where the constraints are no longer algorithmic creativity but physical resources. Silicon. Electricity. Land. Water for cooling. The physics of computation has become the boundary condition for the economics of intelligence.
When you build a data center, you're not just buying chips. You're locking in years of energy contracts. You're securing water rights. You're negotiating with municipal governments for permits. You're building supply chains. The cost is not in the GPU. It's in everything that GPU requires to operate.
$122 billion in the context means: OpenAI can buy the entire physical layer of compute. NVIDIA's GPUs. Land. Power generation. The capacity to train models that smaller competitors cannot even contemplate.
This is not an algorithm war anymore. It's an infrastructure war.
The Four Flywheels of OpenAI
Over the last decade, I've watched companies build moats. I've seen some become monopolies. What OpenAI is building, however, is a unique quad-flywheel system: capital, compute, model, and ecosystem. Each turn of the wheel reinforces the others.
The Capital Flywheel: With $122 billion, OpenAI can outspend every competitor. Anthropic has raised perhaps $10 billion. Google has its own DeepMind, but its structure is a subsidiary, not an independent capital vehicle. The fundraising gap is not a gap. It's a canyon.
The Compute Flywheel: With capital, OpenAI can buy compute in quantities that no one else can match. This means it can train more models, faster, with larger parameter counts. This creates a competitive advantage that compounds each cycle.
The Model Flywheel: Better models attract more users and developers. More users and developers mean more data, more feedback, and more practical improvements. The model improves itself through the ecosystem's interactions with it.
The Ecosystem Flywheel: Developers build on OpenAI's API. Their products become dependent on it. Switching costs become high. The ecosystem becomes a barrier to entry that competitors cannot easily overcome.
Each flywheel is individually powerful. But the combined effect is what makes OpenAI's position nearly unassailable. The competition isn't just about AI anymore. It's about capital markets, energy infrastructure, developer communities, and enterprise sales. Most competitors can match OpenAI on one or two dimensions. None can match on all four simultaneously.
The Contrarian View: Scale Has Limits
I have spent years studying the tension between centralized power and decentralized resilience. My 2020 essay "Code as Covenant" argued that the health of a system depends not on its power but on its constraints. OpenAI's new funding is a test of this principle in an unprecedented context.
There are two hidden vulnerabilities in OpenAI's strategy. First, the compute bottleneck is not just about chips. It's about data. Even with all the compute in the world, models are only as good as the data they are trained on. As synthetic data becomes more common, the quality of the training data becomes a constraint. The internet has a limited amount of text. We're approaching the limits of the "scaling law" that has driven progress since the transformer.
Second, energy is a physical constraint that money cannot easily solve. A single AI cluster can require a gigawatt of power. That's the equivalent of a small city. Building the energy infrastructure to support these clusters is not just a matter of money—it's a matter of regulatory approvals, environmental concerns, and physical construction timelines. Even if OpenAI wants to build data centers, it still needs to wait for the grid to expand, the power plants to be built, and the permits to be issued.
The deeper issue is the acceleration paradox. As we pour more compute into AI, the returns on that compute are diminishing. The easy improvements have been made. The next leap forward—whether it's AGI or something close—may require exponentially more compute for linear improvement. This is not a criticism. It's a physical reality.
Bulls react. Bears reflect. We build. But we build with an understanding of the constraints.
The Competition Landscape
The competitive picture is not just OpenAI versus Anthropic. It's a multilevel structure:
Tier 1: OpenAI with $122 billion. No one else is here.
Tier 2: Anthropic and Google DeepMind with $10 billion to $30 billion in capital. They're strong, but they're in a different league.
Tier 3: Meta's Llama and open-source models. They lack capital but have community and distribution.
Tier 4: The decentralized compute projects like Bittensor, Gensyn, and others. These are small, but they represent an alternative.
The gap between OpenAI and Tier 2 is not a small gap. It's a chasm. The capital advantage will compound over the next few years, making it increasingly difficult for competitors to catch up.
But there's a hidden vulnerability in this dynamic: the closed-source model. OpenAI's strength is also its weakness. By keeping its models closed, it creates an opening for open-source alternatives to gain adoption, especially in areas where developers are concerned about control, transparency, and lock-in.
The open-source ecosystem is not just a competitor. It's a counterweight to centralized AI power. The question is whether open-source AI can survive when the cost of compute becomes prohibitive. If the cost of training the best models reaches $1 billion or $10 billion, then open-source AI will be limited to the elite universities and a few well-funded labs. The community will be left with a "democratized AI" that is a few generations behind.
The Ethical Architecture Question
I write about this issue with a particular bias. My "Soul in the Machine" white paper argued that AI, without an ethical framework, would consolidate power rather than liberate. And the $122 billion raises the question directly: is OpenAI the liberator or the consolidator?
On the one hand, OpenAI has made powerful AI available to the public through ChatGPT. On the other hand, the model itself is a black box. The data is private. The decisions are centralized. And the power is concentrated.
This is not an accusation. It's a structural reality. And the question is: is this acceptable?
For the crypto community, the answer is usually "No." The whole point of decentralization is to distribute power, not concentrate it. If AI becomes a centralized power, then the financial system that runs on AI will be centralized. The data will be centralized. The decisions will be centralized. This is the opposite of the decentralized world we have been building.
But there is a more subtle issue: the alignment problem. In AI, the problem is how to make a superintelligent system that follows human values. In blockchain, the problem is how to make a decentralized system that operates without a central authority. Both are governance problems. Both are about trust.
The difference is that blockchain has a technical solution to the trust problem: code as law. AI has no such solution. It's a much more open problem.
As an AI becomes more powerful, it becomes more dangerous. It can be used for deepfakes, disinformation, cyberattacks, and other malicious purposes. The scale of the damage that it can do is far beyond the scale of the damage that any individual can do.
OpenAI's funding will make these problems more acute. It will create a superintelligence that is more powerful than anything we've ever seen. And it will do so with very little accountability or transparency.
This is not an anti-AI argument. It's a pro-responsibility argument. We need to think about the consequences of this technology, not just the capabilities.
The Global Power Shift: AI, Nation States, and the New Cold War
The $122 billion is not just a private investment. It's a national strategy. The United States is competing with China for AI supremacy. OpenAI is the leading player in this game. And the US government is quietly supporting it.
In the last year, the US has restricted the export of advanced AI chips to China. The message is clear: the US will protect its technological superiority. And the $122 billion is the next step in that strategy.
This has huge implications for the global power balance. It means that the US will control the most powerful AI in the world. It will have the ability to influence other nations through AI, whether it's in the military, economic, or political realm. And it will have the ability to set the standards for AI development.
For the rest of the world, this creates a challenge. They will either have to accept US dominance or build their own AI. The latter will be difficult and expensive. But it may be necessary for national security.
I think about this in the context of decentralized systems. The power of blockchain is that it creates a global, borderless infrastructure that no single nation controls. The AI of the future will be the opposite: a centralized, national infrastructure that is controlled by the US.
This is not a choice between good and evil. It's a choice between different values. And it's a choice that we have to make deliberately, not just accept.
What I See from the Education Platform
I am the founder of a crypto education platform. I have spent years teaching people how to build decentralized systems. I see the AI + crypto convergence as the most important opportunity of our generation.
The convergence is not just about AI agents using crypto to transact. It's about the deep relationship between the two technologies:
- Data provenance: AI models need data. Blockchains can create a verifiable record of where the data comes from and how it's used. This is essential for trust.
- Model verification: AI models are black boxes. Blockchains can create a verifiable record of the model's behavior, making it transparent and auditable.
- Decentralized compute: AI needs massive compute. Decentralized compute networks (DePIN) can provide compute without relying on a single provider.
- Incentive alignment: AI agents can use crypto to create incentive structures that align with the interests of their users.
I believe this is the future. And I believe that the $122 billion investment in OpenAI will accelerate this future, but in a way that may not be healthy. It will create a centralized AI power, and it will create a decentralized response.
The response will be the decentralized AI ecosystem that will rise in opposition. It will be smaller, but it will be independent. It will be less powerful, but it will be more trustworthy. And it will be the place where the values of the decentralized world can survive.
This is the race of the next decade.
The Contrarian Take: The Hidden Costs of Centralized Intelligence
There is a counter-intuitive angle that I want to explore. In the cryptocurrency world, we have a saying: "Not your keys, not your crypto." The same principle applies to AI: "Not your model, not your intelligence."
When OpenAI trains a model, it doesn't just train the model. It creates a representation of the world that is controlled by OpenAI. This representation becomes the basis for how the world sees AI, and how AI sees the world.
This is a subtle but profound power. It's not just about what the model can do. It's about the paradigm of what AI is. The model shapes how people think about intelligence, and how they think about the relationship between humans and machines.
This is the power of "definitive." The model becomes the definition of AI. And that definition is controlled by a single company.
In a decentralized world, the definition of intelligence would be open. It would be contested. It would be evolving. It would be the property of no one. But in a centralized world, the definition of intelligence is a single point of failure.
This is the fundamental tension: the centralization of AI is the centralization of thought. And that is a dangerous thing.
The counterintuitive angle is: the $122 billion investment is not a sign of strength. It's a sign of weakness. It shows that the cost of AI is so high that only a single player can afford it. And this concentration of power is a threat to the AI ecosystem.
The winners of the future will not be the ones who can buy the most compute. The winners will be the ones who can build the most decentralized, the most transparent, and the most trustworthy AI. This will require a new way of thinking about the problem.
The Price of Progress
I want to go back to the beginning of this analysis. Sam Altman said "AI is the most expensive thing." It's a truth that applies not just to the cost of the infrastructure, but also to the cost of the societal impact.
The $122 billion will create the most powerful AI in the world. It will be used for good and for ill. It will solve problems and create new ones. It will be a tool for progress and a weapon for control.
The cost is not just the $122 billion. It's the concentration of power. It's the loss of transparency. It's the risk of misalignment. It's the division between those who have access to AI and those who don't.
Bulls react. Bears reflect. We build.
We build with the understanding that the technology is not the only thing that matters. The values we choose to build are more important.
Tech changes. Values remain.
As I watch the $122 billion investment unfold, I will look not just at the technology, but at the values that underpin it. And I will continue to build a future that is not just more powerful, but more human.
The question is not whether OpenAI will succeed. The question is whether we, as a society, will be able to control the technology we have created.
The answer to this question will determine the future of AI and the future of the world.