In March 2022, the U.S. Office of Foreign Assets Control imposed sweeping sanctions on Russian entities following the invasion of Ukraine. Within 72 hours, Microsoft Azure, AWS, and Google Cloud began cutting off services to affected Russian customers. Businesses that had built mission-critical workflows on those platforms discovered, at speed, that a U.S. government directive could reach inside their operations regardless of where they were headquartered or where their data sat.

That moment was a warning. Most enterprise boards filed it under “geopolitical tail risk” and moved on.
The same logic has now arrived at the AI layer, and the dependencies are deeper, the warning period shorter.

We already have a preview. When Italy’s data protection authority banned ChatGPT over privacy concerns, local businesses relying on it for operational workflows lost access overnight. The Rome Court ultimately annulled the subsequent 15 million euro fine in March 2026, but it did so on a single jurisdictional point: once OpenAI established its Irish subsidiary, the Irish Data Protection Commission became the lead supervisory authority, stripping the Italian regulator of its right to issue a final sanction. The court never examined whether the underlying data practices complied with GDPR. 

Boards should not mistake a jurisdictional escape hatch for an operational green light. The real lesson was the speed of the initial disruption. One regulatory decision, and a core enterprise tool was gone with zero advance notice and zero transition period.

Italy was an early flashpoint. The regulatory landscape has since shifted structurally. The EU AI Act’s high-risk obligations are in the final stages of legislative revision, with a new enforcement deadline of December 2027 agreed in May 2026, a postponement that reflects political complexity rather than reduced intent. The BIS framework in the U.S. is tightening. The question is no longer whether a regulatory action will disrupt your AI operations.

The Dependency You Probably Haven’t Stress-Tested

Ask your CTO a simple question: if your primary frontier AI model provider became unavailable for 30 days due to a regulatory action, an export control directive, or a government-mandated review, what would break, and how quickly?

For most enterprises, the honest answer is deeply uncomfortable. Over the last 18 months, AI has moved far beyond experimental chatbots. It has been woven into autonomous, multi-agent workflows that run core operational pipelines: customer service execution, automated contract analysis, code generation, financial modelling, and compliance screening. When these integrated systems run on a single model, a vendor blackout does not just stall a user query, it halts the automated engine of the business.

Unlike a SaaS CRM or a cloud storage provider, frontier AI models are not commodities. They are concentrated in a handful of U.S.-headquartered companies (Anthropic, OpenAI, Google DeepMind) whose foundational IP and cloud infrastructure are now explicitly subject to tightening U.S. export controls, national security reviews, and data retention mandates that may directly conflict with local privacy regulation. GDPR is only the most obvious example. India’s DPDP Act, Brazil’s LGPD, and the EU AI Act’s transparency requirements all create potential collision points with U.S. vendor terms of service.

Your board does not need to understand transformer architecture. It needs to understand that treating frontier AI as a politically neutral utility, the way you might treat electricity or broadband, is now a critical governance error.

The Strategy: Sovereignty and Hedging

Navigating this requires moving the conversation out of the engineering backlog and into the boardroom, focusing on three strategic pivots.

1. Mandate a Hybrid Model Architecture

The open-weight vs. closed-source debate is no longer an engineering preference; it is a sovereignty conversation. Models like Meta’s Llama series or Mistral can be self-hosted within your own infrastructure perimeter, giving you operational custody and insulation from a foreign vendor’s sudden API kill-switches, executive orders, or unilateral changes to data retention policies.

The right architecture is a tiered model: closed frontier systems reserved strictly for high-stakes, hyper-complex reasoning tasks where capability genuinely justifies the concentration risk; open-weight models running in your own environment for core operational workflows where availability and data sovereignty matter more than the last percentage point of benchmark performance.
The board must demand clear accountability: who owns the decision about which corporate workflows are allowed to tolerate external model dependency, and what is the review cycle?

2. Implement an Independent Orchestration Layer

CFOs do not leave a company’s currency exposure unhedged on the grounds that exchange rates are probably fine. The same discipline should apply to model provider exposure. An intelligent orchestration layer, or model router, must sit between your applications and your model providers. If a primary provider goes offline or changes its terms in ways that conflict with local regulation, the router redirects traffic to a secondary provider or a locally hosted model automatically.

The parallel to treasury is precise: you are not predicting that a provider will fail; you are ensuring that if it does, your operations survive. Do not expect the frontier labs to build this for you. Their business model relies on maximising your consumption of their flagship compute, and they lack your specific business context to route effectively.

This architecture requires planning for graceful degradation. In practice, this means having a fallback ready before you need it. If your primary frontier model goes dark, your orchestration layer must route workflows to a localized, self-hosted model that can securely handle the baseline transaction, keeping core operations running even if advanced reasoning is temporarily unavailable. The cost of building this independent routing layer is a fraction of the operational cost of a 48-hour AI outage across a large enterprise.

3. Map the Fragmented Global Risk Profile

The exposure is not uniform, and a global business must audit its risk based on where its delivery stacks actually sit.

For enterprises with China operations or Chinese ownership structures: This is the one most likely to surface a legal exposure your board does not know it has. U.S. frontier AI models are simply unavailable in mainland China. OpenAI cut off API access in July 2024 following U.S. Treasury restrictions on technology investment flows into China. The risk for global enterprises runs deeper than geography: Anthropic updated its terms of service in September 2025 to prohibit access for any entity more than 50% owned by a company headquartered in a restricted region, regardless of where that entity actually operates. A joint venture with a Chinese majority shareholder, incorporated and operating in Singapore or the UAE, may already be outside the terms of your AI vendor contracts. This is a legal and compliance exposure that needs to be audited now, at the entity level, across your full ownership structure.

For enterprises with significant India operations: India has become the execution layer for global business process automation and autonomous agent deployment. Building those stacks entirely on U.S.-centric closed models imports downstream regulatory risk into every client delivery. Navigating this requires a dual-track strategy. While enterprises must continue to leverage established global models for current production baselines, they must simultaneously fund parallel validation tracks for sovereign alternatives. India’s BharatGen Param2, a 17-billion parameter mixture-of-experts model trained on 22 trillion tokens of multilingual data using government-backed indigenous compute infrastructure, proves that open-weight alternatives are ready for enterprise testing. The immediate mandate for boards is not an immediate shutdown of current APIs, but the funding of shadow testing environments to ensure long-term architecture flexibility.

For U.S.-headquartered enterprises: The regulatory line has been drawn at the computational threshold of 10 to the power of 26 floating-point operations, the statutory boundary establishing a system as a frontier model under California’s Transparency in Frontier Artificial Intelligence Act (SB 53). Developers generating more than 500 million dollars in annual gross revenue face the most intensive obligations under the Act: they must publish annual catastrophic risk frameworks and report critical safety incidents to state emergency agencies within 15 days of discovery, shortened to 24 hours if the incident poses an imminent risk of death or serious physical injury. Violations are enforced by the California Attorney General and carry civil penalties of up to one million dollars per violation. An enforcement action by the California Attorney General against a primary lab would trigger an immediate operational blackout for any single-sourced enterprise. But if your organization also holds federal contracts or operates in regulated markets, that upstream compliance failure will bleed directly into your own legal and audit risk profiles overnight

For European enterprises: The political agreement reached in May 2026 to defer the EU AI Act’s high-risk obligations to December 2027 gives enterprises more runway, but it does not change the architecture decision. The data retention and monitoring policies that U.S. AI vendors operate under remain on a collision course with what European regulation will ultimately require. Using the delay to build compliance-ready infrastructure is the opportunity; treating it as a signal to stand down is the mistake. Domestic alternatives, Mistral and Aleph Alpha, are not inferior substitutes. They are the only providers whose architecture is designed from the ground up to operate within European regulatory constraints.

What Should Be on the Next Board Agenda

Three governance actions, each with a clear executive owner:
Commission an AI dependency audit. Map every workflow that touches an external model provider, classify each by operational criticality, and calculate what a 30-day outage would cost. This risk quantification must produce a concrete number the board can act on.
Assign explicit ownership. Move AI vendor risk onto the enterprise risk register with a named executive owner, likely the CTO or CISO, and a defined quarterly review cadence. If it currently lives nowhere, that gap is itself a governance finding.
Establish a sovereignty threshold. Define what proportion of core operational workflows must run on infrastructure your organisation controls directly, and set a hard timeline for reaching it. This is a strategic policy decision that belongs in the boardroom, not buried in an engineering backlog.

AI is core corporate infrastructure, as consequential to your operational continuity as your ERP or your payments stack. Boards that set a sovereignty threshold now, before an enforcement action forces it, will find that it costs far less to build the architecture than to explain why they didn’t.

It’s not a post I thought I’d be writing in 2026—but this week was one of those weeks where I felt a short refresher might be useful for everyone’s sanity.

Core Principle: Respect the time and cognitive load of the recipient.

When Not to Use Email

 When you are angry or sad: Don’t even risk typing a draft. Just close the app.

 When you need a nuanced or complex discussion: Pick up the phone or hop on a call. Talk it through.

 When it’s a quick, casual check-in: If it can be handled in a single sentence, move it to a messaging platform.

What a Useful Email Actually Looks Like

 A clear subject line: Make it easy to search for and crystal clear on intent (e.g., Action Required, FYI, URGENT). If you need my sign-off, a subject line like ⁠”Need approval for travel to London client meeting”⁠ will get my attention 10x faster than ⁠”Quick question.”⁠

 BLUF (Bottom Line Up Front): Always. Don’t make someone read a three-paragraph thesis before they figure out why you’re writing to them. State the point immediately, offer the explanation context below it, and invite them to chat if they have questions.

 Bold the key takeaways: Let’s face it, very few people read every word of an email. If your message is longer than a couple of sentences, use bold text to guide the reader’s eye to the most critical information.

 Be explicit with the “Ask”: Don’t make the reader guess what their homework is. If you need a response by a certain deadline, say it plainly: ⁠”Need your review of the attached deck by 3 PM ET on 3/7/26.”⁠

 Less is always more: Most people read emails on their phones between meetings. Two paragraphs are usually plenty. Don’t loop in random people just for “visibility,” and please, use Reply All sparingly.

Hot tip: Write the email body before you add the recipients. It is the single easiest way to prevent accidental half-written sends and catastrophic mistakes you will immediately regret.

Three of the most consequential IPO filings in stock market history just happened within two weeks of each other. SpaceX filed its public S-1 on May 20. OpenAI filed confidentially around May 22, targeting a debut above $1 trillion. Anthropic filed its own S-1 on June 1 at a private valuation approaching $1 trillion, with some estimates putting its IPO target as high as $1.75 trillion. If realized, it would be the largest public listing ever attempted.  

The coverage has focused almost entirely on valuation, competition, and whether the hype is justified. Those are reasonable questions. But they are not the most important one. The most important question is not whether these companies succeed. It is what claim the rest of us have on that success, given that the foundations they were built on were not entirely private.

I have pushed back publicly on the doom framing around AI and employment, and I will say it again here. AI is extraordinarily good at specific, well-defined tasks. It is not good at navigating enterprise complexity, managing client relationships under pressure, or making judgment calls in ambiguous situations. Enterprise inertia alone : the sheer difficulty of changing how large organizations actually operate, will keep humans central to most workflows for longer than the headlines suggest. The historical pattern of technology creating more work than it destroys is not broken yet.

But here is what that reassurance misses: You do not need mass unemployment for something to go badly wrong. You only need the normal operation of capital markets.

The Problem the IPO Filings Won’t Mention

By most estimates, NVIDIA’s market cap grew by somewhere in the range of $2 trillion between early 2023 and the end of 2025. Median wages in knowledge work barely moved over the same period. That gap tells the real story. AI is making a small number of companies enormously productive, and the gains are flowing almost entirely to whoever owns the capita : not to the workers, communities, or governments whose infrastructure, data, and publicly funded research made those gains possible.

Consider what AI actually stands on. The transformer architecture underlying most modern AI came out of publicly funded academic research. The internet it runs on was a government project. The data it trained on was generated by billions of ordinary people going about their lives. The power grids it consumes were built with public money and are regulated as public utilities.

This is not an argument for arbitrary wealth redistribution. It is an argument for a return on public investment. The value AI generates from these massive public inputs currently compounds almost entirely in private hands.

Our tools for capturing that value were not built for this. Corporate tax works when companies have local payrolls and physical assets that are hard to move. Large tech companies have spent two decades perfecting the art of moving intellectual property to low-tax jurisdictions. Income tax works when wages grow with productivity. If AI holds wages flat while productivity rises, the tax base stagnates at exactly the moment demand for public services goes up.

The jobs debate gets all the oxygen in the room. The ownership question barely gets asked. That asymmetry is where the real policy failure lives.

What the US Can Actually Do

America built the AI industry. That is a genuine advantage. While government equity stakes in private companies are politically impossible in the US context, the right model already exists closer to home. Since 1982, Alaska has taxed oil companies for using public land and paid every resident an annual dividend. The parallel to AI is direct.

The right mechanism to capture this is a federal levy on AI compute revenue billed within the US. It must target top-line revenue, not profits, which can be re-routed through standard transfer pricing to any low-tax jurisdiction overnight.

A state-by-state approach invites a dangerous race to the bottom, where cloud providers route their next large scale data center expansions to whichever state blinks first. To prevent internal capital flight, the mechanism must be federal, structured as a sovereign wealth contribution to fund a national AI dividend over time.

Predictably, Silicon Valley will argue that taxing compute amounts to unilateral economic disarmament in a critical national security race against state-subsidized adversaries like China. But this presents a false choice. A top-line compute levy does not choke innovation : it funds the very public infrastructure, grid capacity, and research grants required to sustain a long-term technology race. True national security cannot coexist with hollowed-out domestic tax bases.

What India Should Be Doing Differently

I work in this industry, so let me say this plainly: The apocalyptic predictions for India’s technology and services sector, whole industries gone in five years, do not survive scrutiny. Enterprise buying models change slowly. Compliance requirements and hallucination risks keep humans firmly in the loop. Client relationships built over decades carry switching costs that no AI agent can instantly replicate.

But more runway is not the same as no urgency. The FTE-based headcount model is under genuine structural pressure. Companies that treat current stability as permanent will find themselves in a much harder position than those moving now. The direction is clear: from executing processes to redesigning them with AI, from headcount-based billing to outcome-based delivery. It is harder to run, but far more defensible.

On the policy side, India holds an asset the US would struggle to replicate. Aadhaar, UPI, and the Direct Benefit Transfer (DBT) stack form the world’s most capable digital distribution infrastructure. No other country could run a large-scale basic income pilot for workers navigating sectoral transition as cheaply or as quickly. The infrastructure exists. What is missing is the political will to treat this as a serious economic experiment rather than a welfare line item.

India is also leaving its primary negotiating leverage on the table. Global AI models actively seek high-velocity, diverse behavioral and economic data to train localized agents, and currently receive it from Indian citizens entirely for free.

India must stop approaching these negotiations with a 20th-century “manufacturing plant” mindset, content with mere construction jobs and local real estate investment. The goal should be aggressive data sovereignty and revenue-sharing arrangements. If global AI companies want to build data centers here and integrate with our digital public infrastructure, India should be negotiating for sovereign equity stakes, not just tax receipts.

The Honest Summary

The job destruction story is overblown. The wealth concentration story is underreported. Conflating them produces the wrong policy conversation.

Preparing for AI does not mean panicking about unemployment. It means building the ownership structures and revenue-sharing mechanisms before the value fully concentrates, because redistributing wealth after it has concentrated is an order of magnitude harder than claiming a fair share of it upfront.

The wealth from AI is real and it is compounding now. The question is whether we build the pipes to share it before the ownership patterns harden into something permanent, or spend the next decade having an argument about jobs while the more consequential decisions get made without us.