The silence was the first thing I noticed. In the flood of analyses following OpenAI's $6.4 billion acquisition of io, Jony Ive's hardware startup, one voice was conspicuously absent from the initial chorus. Dario Amodei, the CEO of Anthropic, didn't tweet. He didn't issue a press release. Instead, a carefully worded report surfaced, almost buried in the noise, suggesting he had proposed a 'three-step strategy' for responsible AI development, one that hinged on global cooperation.
Question the whisper. When the loudest players are silent on a major industry event but quietly leak a governance framework, the alpha isn't in the acquisition price. It's in the silence of the audit that follows.
I spent three days digging into the technical filings, the governance proxies, and the public records of both companies. What I found wasn't just a story about a hardware deal. It was a story about the emerging 'Safety-Industrial Complex,' where the definition of 'responsible AI' is becoming the most valuable asset in the $600 billion AI race. Amodei's three-step whisper isn't a plea for global harmony. It is a strategic move to make Anthropic's internal safety protocols the de facto global standard, turning compliance into a competitive moat that only the wealthiest players can afford.
Context: The Protocol Behind the Whisper
To understand why Amodei's three-step strategy is a narrative shift, you have to look at the history of AI governance cycles. We have moved from the 'Ethics Committee' era of 2018-2020, characterized by academic hand-wringing and vague principles, to the 'Regulatory Hard Fork' era of 2023-2025. The EU AI Act is now law. The US Executive Order on AI has mandated reporting requirements for models trained above 10^26 FLOPS. The 'move fast and break things' narrative is dead.
Dario Amodei is not a politician. He is the CEO of Anthropic, a company valued at upwards of $60 billion, whose core differentiation from OpenAI has never been raw model capability. It has been safety. Their 'Responsible Scaling Policy' (RSP) isn't a PR document; it's a technical framework that classifies models by 'AI Safety Level' (ASL), similar to biosafety levels. It dictates that as models approach certain capability thresholds—like the ability to autonomously replicate or conduct cyber operations—specific security and deployment protocols must be activated.
The three-step strategy Amodei proposed, though its specifics remain frustratingly opaque, aligns perfectly with a governance sentiment analysis of the current market. The steps, as inferred from his public testimony and Anthropic's RSP, likely follow this logic:
- Transparency Mechanisms: Mandating that frontier labs publish standardized safety evaluations, similar to financial audits. This is the 'Read the docs' phase applied to model capabilities.
- Compute Governance: Implementing international verification regimes for large-scale training runs. This is the technical enforcement mechanism, a way to physically track the 'talent' (computational power) required to build AGI.
- Constitutional Alignment: A global agreement on the fundamental principles (a 'constitution') that governs AI behavior, moving beyond vague ethics to enforceable code-level constraints.
This is not a utopian vision. This is a blueprint for a regulated market. And that, for a fund manager, is the most important signal.
Core Analysis: The Safety-Industrial Complex and the Economics of Trust
My analysis of the governance sentiment around this proposal reveals a stark reality: The 'three-step strategy' is a masterclass in turning a regulatory burden into a strategic asset.
Let's look at the data. In 2025, the cost of training a frontier model exceeds $100 million. The compute required is concentrated in the hands of a few hyperscalers. Amodei's strategy, if implemented, would add a layer of compliance costs—audits, reporting, verification—that would disproportionately impact smaller labs and open-source projects.
This is not a hypothetical. It is the same pattern we saw with MiCA in Europe. The regulation was touted as 'clarity,' but the operational reality—the reserve requirements for stablecoins, the capital requirements for CASPs—created a barrier to entry that favored incumbents like Circle and Coinbase. The 'clarity' was a killer for small projects.
The same dynamic is now playing out in AI. Consider the 'Trust & Ethics' score I now apply to every investment thesis. For AI companies, the metric is shifting from 'Benchmark Performance' to 'Governance Maturity.' If Amodei succeeds in establishing a global norm where 'safety' is defined by a specific, auditable standard (likely modeled on Anthropic's RSP), then Anthropic's compliance costs are fixed. They already built the infrastructure. For a competitor like Mistral or a decentralized AI project, the cost of retrofitting that level of safety infrastructure could be existential.
This is the 'Safety-Industrial Complex' in action. It is a sociotechnical feedback loop where the definition of safety is written by the players who can afford to be safe. The 'global cooperation' narrative is the sugar that makes the medicine go down. It appeals to the public good, but its technical implementation creates a centralizing force.
Take the 'compute governance' step. How do you verify a 10^26 FLOPS training run? You need hardware-level attestation, supply chain transparency, and international monitoring. This is a technical mandate that only the largest cloud providers (AWS, Azure, Google Cloud) and hardware manufacturers (Nvidia) can realistically enforce. It effectively turns private infrastructure into a regulatory apparatus. For a fund manager, this means the real value isn't just in the model creators. It's in the picks and shovels of the audit layer: the companies building verifiable compute, the zero-knowledge proof systems for model inference, and the third-party auditors who will become the 'Big Four' of AI safety.
Alpha hides in the silence of the audit. Amodei isn't just talking about safety. He is carving out the regulatory perimeter that protects his market position while forcing competitors to play by his rules.
Contrarian Angle: The Global Cooperation Illusion and the Sovereignty Paradox
The prevailing narrative in the mainstream crypto and tech media is that Amodei's proposal represents a noble, if ambitious, attempt to prevent an AI catastrophe. The contrarian view, the one that is being whispered in the corridors of Brussels and Washington, is that this 'global cooperation' framework is fundamentally incompatible with the geopolitical reality of 2026.
We are not in a period of global cooperation. We are in a period of techno-nationalism. The US is tightening chip export controls. China has responded with its own 'Global AI Governance Initiative,' which explicitly rejects the hegemony of Western-led safety standards. The three-step strategy, by advocating for a single global standard, is not a bridge between these blocs. It is a bid to make the Western, private-led standard the only legitimate one.
My experience running investor counseling during the FTX collapse taught me a hard lesson about trust in centralized narratives. Trust is the most scarce asset in crypto, and it is about to become the most scarce asset in AI. The 'three-step strategy' is a play for that trust. But here's the blind spot: If the strategy is seen purely as a US-led corporate initiative, it will fail in the Global South and in China. The risk is not that it doesn't work. The risk is that it creates a 'splinternet' of AI safety.
We could end up with a bifurcated AI industry: a 'high-compliance' bloc aligned with US standards (Anthropic, OpenAI, Google, and their European partners) and a 'sovereign' bloc (China, and potentially other nations) that uses different safety protocols. For investors, this is a nightmare scenario because it fragments liquidity, doubles compliance costs, and creates massive regulatory arbitrage opportunities for bad actors.
In my 2024 essay series, 'From Speculation to Sovereign Reserve,' I argued that the Bitcoin ETF was an educational tool for institutional mothers. This is different. The AI governance framework Amodei proposes is not an educational tool. It is a control rod. And if it is not designed with genuine multi-stakeholder input—not just from US tech CEOs but from international bodies that include competing powers—it will be rejected as a tool of cultural and economic imperialism.
The three-step strategy, for all its technical precision, lacks a fourth step: a mechanism for legitimate, neutral dispute resolution. Without it, it's just a code of conduct for a club of billionaires.
Takeaway: Trading the 'Trust Infrastructure' Layer
So, how do we position for this narrative shift?
The shift from 'Capability' to 'Governance' as the key competitive differentiator is the most significant macro trend in AI for the next 24 months. It means the investment thesis for AI cannot just be about model architectures or user growth. It must include a 'Trust & Ethics' audit of the governance stack.
The next narrative is not about who builds the smartest AI. It is about who builds the most transparent, verifiable, and compliant AI. The alpha will be found in the 'Trust Infrastructure' layer: the protocols providing zero-knowledge proofs for model inference, the decentralized identity systems for AI agents, and the hardware-based attestation solutions that will make 'compute governance' a reality.
As I look at the freshly funded AI-crypto hybrid projects raising $100 million on the promise of 'decentralized intelligence,' I apply my due diligence checklist. I ask: What is their 'Constitutional AI' framework? How do they handle a 'three-step' style audit? Who audits the auditors? If the answer is silence, the risk is not technological. It is regulatory and ethical.
Amodei has fired a shot across the bow of every AI founder. He has signaled that the era of unbridled scaling is over. The era of audited scaling has begun. The question for us, as investors and as citizens of the digital world, is not whether we can afford to cooperate. It is whether we can afford not to. In the silence of the audit, the future of the industry is being written. We must read the docs very carefully.
