The math doesn't lie, but it can be stretched. Last week, a report surfaced claiming Anthropic projects $190–$200 billion in revenue by 2028. The source: unnamed insiders. The method: revenue multiples. The implication: a valuation north of $2 trillion. For a company that didn't exist five years ago, this is not just ambitious—it is an institutional declaration of war on traditional software economics. But as a DeFi security auditor, I see a different story. Not about AI, but about the infrastructure that will either support or collapse under such a weight. And for the crypto ecosystem, this is a signal that cannot be ignored.
Context: The Numbers and the Narrative
Anthropic, the AI safety-focused company behind Claude, is reportedly telling investors to expect $190–$200 billion in annual revenue by 2028. This is not a typo. To put it in perspective, Microsoft's entire fiscal 2024 revenue was ~$245 billion. Google's was ~$350 billion. Anthropic is claiming that within seven years of founding, it will be a top-20 global software company by revenue. The justification? A combination of enterprise AI adoption, agentic workflows, and a pricing model that assumes continuous technological superiority.
The report cites four anonymous sources who claim bankers and investors are using enterprise value-to-revenue multiples to value the company. This is standard for high-growth, pre-profit firms. But the twist is the forecast horizon: three years out. In traditional SaaS, you rarely see revenue projections beyond 12–18 months because the market is too volatile. Here, the market is accepting a 2028 target as a credible anchor. This is a departure from rational pricing. It is narrative pricing.
From my experience auditing DeFi protocols, I recognize this pattern. When a project starts using forward-looking metrics that are not supported by current data, it is often a warning sign of over-leverage. The same applies to AI companies. The difference is that Anthropic has real revenue—$47 billion annualized as of May 2025—and a clear path to growth. But the jump from $47B to $200B requires a CAGR of ~60%. That is aggressive, even for a market that is doubling every year.
Core: Code-Level Analysis of the Infrastructure Assumptions
Let me break down what $200 billion in revenue actually means in terms of infrastructure. This is where my background in protocol security becomes relevant. Every dollar of AI revenue requires compute, storage, and network resources. The cost structure is not like software where marginal cost is near zero. AI inference is expensive. At current token prices, a $200B revenue target implies inference costs of $60–$80 billion annually, assuming 60–70% gross margins. That is a massive capital expenditure.
To achieve that, Anthropic would need to reduce per-token inference costs by 5–10x over three years. How? Through custom silicon, better model architecture, and economies of scale. But here is the catch: the timeline for custom ASIC development is 18–24 months minimum, and that is if you have a team of engineers with deep experience. Anthropic is reportedly working with chip designers, but public details are scarce. In my experience, when a company's revenue target depends on unproven hardware, the risk is asymmetric. If the chip delivers, great. If it doesn't, the entire model fails.
Now, consider the crypto angle. The compute required for $200B in AI revenue is on par with the entire Bitcoin mining network's energy consumption. According to the Cambridge Bitcoin Electricity Consumption Index, Bitcoin mining uses ~150 TWh annually. Anthropic's inference demands could be similar or higher. This means that the company will need to secure long-term power purchase agreements, build data centers, and potentially tap into decentralized compute networks. This is where blockchain projects like Akash, Render, or Golem could play a role. But the scale is so large that these networks would need to grow by orders of magnitude. Currently, decentralized compute is a fraction of the market. It is not ready for prime time.
I have audited several decentralized compute protocols. The most common vulnerability is trust in the node operator. If you are running a model that costs $10,000 per inference hour, you cannot afford to have a rogue node returning incorrect results. The verification mechanisms—replication, cryptographic proofs, or auditing—are either too slow or too expensive. For Anthropic to rely on decentralized compute, the protocols would need to be redesigned from the ground up. That is not happening by 2028.
Contrarian: The Blind Spot of Infinite Growth
The contrarian take is not that Anthropic will fail. It is that the market is pricing in a scenario where AI growth is unbounded by physical constraints. The Silicon Valley mantra is that software eats the world. But AI is not just software. It is software plus hardware plus energy. The energy constraint is real. The chip constraint is real. The talent constraint is real.
Consider the supply chain. The world's leading chip foundry, TSMC, is already at capacity. Nvidia's H100 and B100 GPUs are sold out for months. If Anthropic wants to build a custom ASIC, it needs to compete for wafer starts with Apple, AMD, and Nvidia. That is a zero-sum game. The same applies to power. In the US, the grid is already strained. Data center power demand is expected to double by 2030. Anthropic's expansion alone could require multiple gigawatts of capacity. That means building new power plants, transmission lines, and substations. Permitting alone takes 5–10 years.
Now, the crypto blind spot: many in the blockchain space assume that AI will drive demand for decentralized compute. But the opposite is more likely. Large AI companies like Anthropic will prefer centralized, vertically integrated infrastructure because it offers reliability, security, and control. They will not trust open networks for mission-critical workloads. The only exception is if the decentralized network offers a cost advantage that is significant enough to overcome the trust deficit. But given the scale of $200B revenue, the cost savings would need to be in the tens of billions to justify the added complexity. That is unlikely.
Takeaway: The Vulnerability Forecast
So what does this mean for the crypto industry? First, the AI narrative is a double-edged sword. On one hand, it attracts capital and talent to compute-intensive projects. On the other hand, it creates unrealistic expectations. The $200B target is a signal that the market is willing to accept extreme valuations based on forward-looking narratives. This is reminiscent of the ICO mania in 2017, where projects promised billions in revenue without a clear path to delivery. The difference is that Anthropic has real product-market fit. But the infrastructure is not ready.
My forecast: within the next 18 months, we will see a major AI infrastructure bottleneck. Either energy prices will spike, or chip supply will constrain growth, or both. When that happens, the narrative will shift from 'infinite growth' to 'sustainable growth.' The crypto projects that focus on real-world compute efficiency and reliability will survive. The ones that rely on hype will fail.
Trust the code, verify the trust. And in this case, the code is the infrastructure. Until I see a detailed plan for how Anthropic will source the compute, energy, and chips to support $200B in revenue, I remain skeptical. The math doesn't lie, but it often hides the cost of truth.