Three regimes of capital — sovereign wealth, the venture-intelligence model, and regulation itself — are reshaping the innovation frontier in different directions, and Europe’s least-recognised instrument is a regulatory one

It is a commonplace that money follows innovation — that capital, sensing opportunity, flows toward the frontier where returns are highest. The commonplace is true, and it is also, for the purposes of strategic analysis, nearly useless, because it describes capital as a passive follower of a frontier defined elsewhere. The more consequential truth runs the other way. Capital does not merely follow the frontier; it shapes it. The form capital takes — who supplies it, on what terms, with what objective, subject to what constraint — determines which technologies get built, which get abandoned, and which never reach the threshold of seriousness at all. The frontier is not a fixed landscape that money discovers. It is, in significant part, a product of the money that funds it.

If that is right, then the strategically important question is not how much capital is flowing toward emerging technology — a great deal is, and the figure changes weekly — but what kind of capital, structured how. This essay argues that three structurally distinct regimes of capital are currently shaping the innovation frontier, that each shapes it in a characteristic and different direction, and that they should be analysed as three different instruments rather than as three sources of the same undifferentiated thing. The first is sovereign wealth: state-owned capital deployed at scale as an instrument of statecraft. The second is the venture-intelligence model, of which the CIA-chartered In-Q-Tel is the archetype: patient, mission-directed capital that uses the discipline of the private market to serve a public objective. The third is the least recognised and, this essay contends, the most underrated — regulation itself, functioning as a form of capital, of which the European Union is the principal and largely unwitting master.

The argument is not that these three exhaust the field; conventional venture capital, corporate research budgets, and direct state procurement all remain enormous and consequential, and the companion analysis of the defence-technology markets elsewhere in this volume treats the venture layer at length. The argument is that these three regimes are distinctive in a specific way: each converts capital into strategic advantage through a different mechanism, and understanding the mechanism is the key to anticipating the direction in which each will bend the frontier. The three are treated in turn, with the sovereign-wealth and venture-intelligence cases grounded strictly in the public record, and the regulation-as-capital argument — which is this essay’s original contribution — developed as an analytical proposal and defended against its strongest objection.

I. Sovereign wealth: capital as statecraft

The first regime is the most visible and the most straightforwardly enormous. Sovereign wealth funds — state-owned investment vehicles, historically built on commodity surpluses — have become, in the past several years, among the most active and directional pools of capital in the world, and their deployment into frontier technology is increasingly indistinguishable from an instrument of foreign policy. The scale is difficult to overstate. The Gulf funds alone — Saudi Arabia’s Public Investment Fund, Abu Dhabi’s ADIA, Mubadala and ADQ, Qatar’s QIA, and Kuwait’s KIA, collectively the ‘Oil Five’ or, with their affiliates, the main Gulf seven — collectively manage on the order of five trillion dollars, and in 2025 the seven largest accounted for something approaching half of all capital deployed by state-owned investors globally, a record concentration.

What matters for this essay is not the scale as such but the direction and the logic. This capital is flowing toward the frontier deliberately and strategically. Sovereign-owned investors put on the order of sixty-six billion dollars into artificial intelligence and digitalisation in 2025, with the Gulf funds leading; Abu Dhabi’s Mubadala alone invested nearly thirteen billion dollars in AI and digital that year. The vehicles are increasingly purpose-built: MGX, the AI-focused fund co-founded in 2024 by Mubadala and the Emirati AI holding company G42, launched with a hundred-billion-dollar orientation toward AI infrastructure, semiconductors, and core AI technologies, and has since participated in data-centre transactions of a scale — a reported forty-billion-dollar acquisition alongside a major infrastructure investor — that only sovereign-scale capital could contemplate. Saudi Arabia’s PIF, at roughly nine hundred billion dollars under management and targeting two trillion by 2030, has made frontier technology and compute central to its Vision 2030 restructuring.

The characteristic feature of sovereign-wealth capital — the thing that makes it a distinct regime rather than merely large conventional investment — is that its objective function is not primarily financial. It is what one analytical tradition has usefully termed instrumental capital: money deployed to achieve political and strategic ends, with financial return a constraint rather than the goal. A sovereign fund that builds a domestic AI ecosystem, secures preferential access to frontier compute, or acquires a position in a strategically important foreign technology is engaged in statecraft conducted through an investment vehicle. The investment is the diplomacy. This is why the Gulf funds’ technology deployments are inseparable from their states’ broader strategies of economic diversification, strategic hedging between the United States and China, and the conversion of a wasting hydrocarbon endowment into durable technological and geopolitical relevance.

The direction in which this regime bends the frontier follows from its logic. Sovereign-wealth capital favours the capital-intensive, infrastructure-heavy, strategically legible end of the frontier — data centres, compute, semiconductors, foundational AI — because these are the assets that confer sovereign capability and geopolitical position, and because their very scale suits capital that need not justify itself on a venture timeline. It is patient where venture capital is impatient, directional where venture capital is diversified, and political where venture capital is financial. It builds the physical substrate of the frontier. What it is structurally less suited to fund is the early, uncertain, high-failure-rate discovery from which genuinely novel capability emerges — the domain of the second regime.

Epistemic status: the sovereign-wealth figures are Confirmed from public reporting current to late 2025 and early 2026, and are point-in-time; assets under management, deal values, and fund strategies move quickly and should be refreshed at publication. The characterisation of sovereign-wealth capital as ‘instrumental’ — deployed for strategic rather than primarily financial ends — is an established analytical frame, not the essay’s invention, though the application to the frontier is developed here.

It is worth distinguishing the Gulf model from two other species of sovereign capital, because the differences bear on how each shapes the frontier. Norway’s fund — the largest in the world by some measures — is the archetype of the passive, return-maximising sovereign investor: broadly diversified, transparency-bound, and deliberately insulated from statecraft, it is sovereign capital that has chosen not to be instrumental, and it shapes the frontier hardly at all beyond its sheer weight as a shareholder. Chinese state-directed capital sits at the opposite pole: fused with industrial policy and the state’s technological priorities, deployed through vehicles whose objectives are inseparable from the party-state’s, it is instrumental capital in which the financial and the strategic are not merely linked but identical, as the companion analysis of civil-military fusion in this volume describes. The Gulf funds occupy a distinctive middle position — more instrumental than Norway, more commercially disciplined and more internationally entangled than China’s vehicles, hedging between the great powers rather than serving one. The three are all ‘sovereign wealth’, and they shape the frontier in three different ways, which is a further reason the undifferentiated category is analytically inadequate.

II. The venture-intelligence model: patient capital with a mission

The second regime is smaller by orders of magnitude and disproportionately influential, and it is best understood through its archetype. In-Q-Tel, chartered by the Central Intelligence Agency in 1999 as an independent, non-profit venture-capital vehicle, was created to solve a specific problem: the government’s procurement machinery was too slow and too insular to keep pace with a private technology sector that had become the primary engine of innovation. The solution was to build a venture-capital firm that would invest, on the government’s behalf, in early-stage companies developing technologies of intelligence and national-security relevance — and, crucially, to do so through the disciplines of the commercial market rather than the mechanisms of government funding. By 2025, In-Q-Tel had reached its eight-hundredth investment, and its documented portfolio over the years has included companies — Palantir among the best known — that became significant in their own right.

The model’s distinctive features are worth stating precisely, because they define the regime and because the temptation to romanticise In-Q-Tel is strong and should be resisted. First, its objective is not to maximise financial return. Its purpose is to accelerate the development and delivery of specific technological capabilities to its government customers, with financial return a secondary consideration and, in effect, a discipline that keeps its portfolio companies subject to real market pressure rather than a goal in itself. Second, it exploits a leverage that conventional venture capital lacks: the government as an anchor customer and beta-tester. A young company in the In-Q-Tel portfolio gains not only capital but a demanding, credible early user whose requirements shape the product and whose adoption de-risks it for the broader market. Third, it is a bridge institution by design — its function is to connect the pace of the private frontier to the needs of a public mission, translating between two systems that operate on very different clocks.

The direction this regime bends the frontier is distinct from the sovereign-wealth case and, in a sense, complementary to it. Where sovereign wealth builds the frontier’s expensive physical substrate, the venture-intelligence model reaches toward its early, uncertain, discovery-heavy edge — the small companies and unproven technologies where a modest, well-directed investment paired with a credible customer can pull a capability across the threshold from research into use. It is a mechanism for converting the government’s knowledge of its own needs into a signal that shapes what the private frontier produces. Its scale is small; its directional influence, because it operates at the point where technologies are still malleable, is large relative to that scale.

Honesty requires stating the limits and the criticisms, and the plan for this essay rightly insists that the In-Q-Tel claims be held to the public record. The model is not a costless success story. Scholarly and legal analysis has questioned whether an intelligence agency is well suited to the notoriously difficult business of venture capital, has raised ethical concerns about government-sponsored equity investment in private companies, and has noted that a firm investing in companies with international operations can expose its sponsoring agency to entanglements it would otherwise avoid. There are also structural questions about attribution and selection that the public record cannot fully resolve: it is genuinely difficult to establish, from outside, how much of a portfolio company’s success is attributable to the In-Q-Tel relationship rather than to factors that would have obtained regardless. This essay therefore makes the narrow claim the evidence supports — that the venture-intelligence model is a distinctive regime that shapes the early frontier through mission-directed, customer-anchored patient capital — and declines the broader claim, unsupported by public evidence, that it is straightforwardly the most effective of the three.

Epistemic status: In-Q-Tel’s charter, non-profit structure, congressional funding, the 2025 eight-hundredth-investment milestone, and its documented portfolio are Confirmed from the public record. The model’s internal performance, the counterfactual value of its investments, and any classified dimension of its activity are not publicly establishable, and the essay’s claims are deliberately confined to what the open record supports. A specialist reviewer with venture or national-security-investment experience is recommended for this section, per the essay brief.

The venture-intelligence model has, tellingly, generalised, which is evidence that it captures something real rather than a one-institution idiosyncrasy. The United States Department of Defense established the Defense Innovation Unit explicitly to piggyback, in a former defence secretary’s own word, on the work the intelligence community had done through In-Q-Tel — extending the bridge-institution logic from intelligence to the broader military. Allied and multilateral versions have followed the same template: NATO’s innovation fund and the accompanying accelerator programme apply the model at the alliance level, deploying patient, mission-directed capital into early-stage deep-technology companies of defence relevance across member states. The proliferation matters for the argument because it demonstrates that the regime is a transferable instrument, not an artefact of one agency’s history: wherever a government concludes that its procurement machinery cannot keep pace with the private frontier, some version of the venture-intelligence model tends to appear, because the underlying problem — the mismatch between the clock of government and the clock of the frontier — is general.

III. Regulation as capital: Europe’s unrecognised instrument

The third regime is the essay’s central and most counter-intuitive claim, and it requires a conceptual move before it can be argued. The claim is that regulation — ordinarily understood as a constraint on economic activity, a cost imposed on innovation, the opposite of capital — can function as a form of capital, and that the European Union, widely regarded as having fallen behind the United States and China in the frontier-technology race precisely because it over-regulates, in fact commands through its regulatory power an instrument of frontier-shaping influence that its critics, and to a considerable extent the EU itself, have failed to recognise as such.

Begin with the conceptual move. Capital, in the sense that matters for shaping the frontier, is anything that can be deployed to determine which technologies get built and on what terms. Money is the obvious form. But access to a market of four hundred and fifty million relatively wealthy consumers is also something that can be deployed to determine which technologies get built — because a firm that wishes to sell into that market must build its products to the market’s specifications, and whoever sets those specifications thereby exercises a shaping power over the product that is functionally equivalent to the shaping power of an investor who attaches conditions to a cheque. The EU does not need to fund a technology to shape it. It needs only to set the terms on which the technology may be sold to its citizens, and the size and wealth of its market do the rest. This is regulation functioning as capital: not money spent, but market access conditioned, to the same frontier-shaping effect.

The mechanism by which this works is well documented, and it is not this essay’s invention; the essay’s contribution is to reframe it as a species of capital rather than merely a species of influence. The legal scholar Anu Bradford named the phenomenon the Brussels Effect: the EU’s capacity to set global standards not through coercion or treaty but through the appeal of its market, such that multinational firms, finding it more efficient to build a single compliant product than to maintain separate compliant and non-compliant versions, adopt the EU’s requirements globally. The General Data Protection Regulation is the canonical case: the world’s websites reorganised their handling of personal data around a European rule, not because they were compelled to but because the cost of serving the European market on any other terms was higher than the cost of compliance. The AI Act, with its conformity-assessment requirements for high-risk systems, is the EU’s attempt to repeat the manoeuvre in artificial intelligence, complemented by a de jure effect through which jurisdictions lacking their own regulatory capacity adopt the European template wholesale.

Reframed as capital, the Brussels Effect looks quite different from the way it is usually discussed. The usual debate asks whether European regulation helps or hurts European competitiveness — whether the AI Act will strangle a nascent European AI sector or protect European citizens, whether the EU regulates because it cannot innovate. That debate treats regulation as a cost. The regulation-as-capital frame asks a different question: what frontier-shaping power does the EU exercise by setting the terms of access to its market, and is that power being deployed strategically or squandered? On this view, the EU’s regulatory apparatus is a capital instrument of formidable reach — an instrument that can shape the global development of a technology without the EU’s needing to fund, build, or own any part of it — and the strategically relevant failure is not that Europe regulates but that it does not recognise its regulation as the frontier-shaping instrument it is, and therefore does not wield it with the intentionality that a sovereign-wealth fund brings to its deployments of money.

The direction this regime bends the frontier is characteristic and, properly understood, powerful. Regulation-as-capital shapes the frontier toward the values and requirements the regulation encodes — privacy, transparency, safety, human oversight, whatever the regime specifies — by making those requirements a condition of market access and letting the market’s size propagate them globally. It is capital that shapes not by choosing which technologies to fund but by setting the terms every technology must meet, and its reach is potentially wider than any fund’s, because it operates on all products seeking the market rather than on the subset an investor selects. For a bloc that cannot match American venture depth or Gulf sovereign scale, this is not a consolation prize. It is, this essay contends, Europe’s genuine and underexploited comparative advantage in the shaping of the frontier: the one instrument on which Europe leads the world, hidden in plain sight and mislabelled as a weakness.

Epistemic status: Probable, and this is the essay’s original analytical contribution. The Brussels Effect itself — the EU’s market-driven global standard-setting through GDPR and the AI Act — is Confirmed and well documented in the scholarly literature. The reframing of that effect as a species of capital, and the consequent claim that regulation-as-capital is Europe’s underrated frontier-shaping instrument, is the author’s analysis, offered as a strong argument rather than an established finding. It is defended against its principal objection immediately below.

IV. The case against, and why the argument survives it

The strongest objection to the regulation-as-capital thesis is not that regulation is a cost rather than capital — that objection mistakes the claim, which is precisely that the two framings describe the same instrument from different angles. The strongest objection is empirical and specific: that the Brussels Effect, on which the whole argument rests, is weaker for artificial intelligence than it was for data protection, and may not hold at frontier scale at all. This objection is serious and must be met rather than waved away.

The objection runs as follows. The Brussels Effect depends on a particular market structure. It works when a product is non-divisible — when it is more efficient for a firm to build one globally compliant version than to fork its product and maintain two — and when the EU can credibly threaten to exclude non-compliant products from a market too valuable to abandon. For data protection, both conditions held: reorganising data handling globally was cheaper than maintaining two regimes, and no major platform would forgo the European market. For artificial intelligence, analysts at Brookings and in the standard-setting literature have argued, the conditions hold more weakly. The AI market is more fragmented and application-specific than the platform market; large exporters already selling into the EU will lobby to shape the rules rather than accept them passively; and firms may find it easier to fork AI products — to offer a compliant version in Europe and a different one elsewhere — than they did to fork their data practices. On this account, the AI Act will have real extraterritorial influence but will not straightforwardly become the global standard, and the regulation-as-capital instrument is therefore weaker than the essay claims.

Concede the force of this, because it is correct as far as it goes, and the concession sharpens rather than defeats the argument. It is true that the Brussels Effect is not automatic and not uniform — that it applies strongly to some categories of product and requirement and weakly to others, and that AI’s market structure makes it a harder case than data protection. But three considerations preserve the thesis. First, ‘weaker than for GDPR’ is not ‘absent’; the scholarly consensus is that a de facto effect is likely for a meaningful subset of high-risk AI systems and a de jure effect likely among jurisdictions lacking their own regulatory capacity, which is a substantial frontier-shaping reach even if it falls short of universal. Second, the objection is an argument about the current strength of one instance of the instrument, not about the validity of the frame: even a partial Brussels Effect is regulation functioning as capital, and the analytical point — that the EU should recognise and wield this instrument strategically — stands regardless of whether the AI Act specifically achieves GDPR-scale propagation. Third, and most important, the objection actually reinforces the essay’s central practical claim. If the effect is not automatic, then whether it materialises depends on how deliberately the instrument is deployed — on whether the EU sets its standards with the intentionality of an investor structuring a deal, times them to the moment of maximum leverage, and coordinates them across its regulatory apparatus. An instrument that works automatically requires no strategy; an instrument that works only when wielded well requires exactly the strategic recognition the essay argues Europe lacks. The objection, pressed to its conclusion, is an argument for taking regulation-as-capital more seriously, not less.

V. Three instruments, three frontiers

Set the three regimes side by side and the analytical payoff of treating them as distinct instruments becomes clear. Sovereign wealth shapes the frontier by building its capital-intensive substrate, deploying instrumental capital at a scale only states command, and bending the frontier toward the assets that confer sovereign capability — compute, infrastructure, foundational technology. The venture-intelligence model shapes the frontier by reaching its early, uncertain edge, pairing modest mission-directed capital with the leverage of a credible government customer, and pulling specific capabilities across the threshold from research into use. Regulation-as-capital shapes the frontier by setting the terms every product must meet to reach a vast market, propagating the values it encodes globally without funding or owning anything, and bending the frontier toward whatever the regulation specifies. Three instruments, three mechanisms, three directions.

The strategic reading that follows is that a serious power does not choose among these instruments; it understands which it commands and wields each for what it can do. The United States commands the deepest conventional venture market and, through the venture-intelligence model, a distinctive mechanism for directing early discovery; its challenge is coordination across a system that is powerful but diffuse. The Gulf states command instrumental capital at sovereign scale and are deploying it with increasing intentionality; their challenge is converting purchased position into indigenous capability. And the European Union commands the regulation-as-capital instrument more completely than any other actor commands any of the three — and is the least aware that it holds a frontier-shaping instrument at all, persisting in the belief that its regulatory power is the mark of its technological failure rather than the form of its residual technological influence.

One further observation completes the picture, because the three regimes do not operate in separate compartments; they increasingly compete for the same scarce assets, and the competition is itself frontier-shaping. Sovereign wealth, venture capital, and the venture-intelligence model are all now bidding for positions in the same small set of frontier-technology companies and the same constrained supply of compute, and the terms on which they bid differ in ways that matter. Sovereign capital can outlast and outspend; venture capital can move faster and tolerate more failure; mission-directed capital can offer the one thing neither of the others can, a credible government customer. A frontier company choosing among them is choosing not merely a valuation but a shaping influence, and the regulation-as-capital instrument sits over all of them, setting the terms any resulting product must meet to reach the European market. The frontier that emerges is the joint product of this competition — which is the deepest sense in which the money, and the rules the money must obey, shape the frontier rather than merely following it.

That last point is where the essay’s argument bears most directly on strategy, and it is the claim worth ending on. The dominant European self-understanding — shared by many of Europe’s critics and too many of its own officials — is that the continent has lost the frontier-technology race and that its regulatory reflex is a symptom of that loss. The regulation-as-capital frame inverts this. Europe has not been reduced to regulating because it cannot innovate; it possesses, in its regulatory power, a genuine instrument for shaping the global frontier that neither the United States nor China can match, and it has failed to recognise and deploy that instrument with the strategic intentionality that its rivals bring to their money. The money shapes the frontier. So does the rule that money must obey. Europe’s most underrated defence instrument is a regulatory regime — and the first step to wielding it is to see it, at last, as capital.

References

Sources for this essay, grouped by capital regime. Current to late 2025 / early 2026; figures are point-in-time.

Sovereign wealth

Global SWF, data on 2025 sovereign-investor activity (the Gulf seven accounting for ~43% of state-owned investor capital; ~$66bn into AI and digitalisation; PIF as the single largest dealmaker of 2025).

Stiftung Wissenschaft und Politik (SWP), “Sovereign Wealth Funds and Foreign Policy,” 2026 (the ‘Oil Five’; sovereign capital as an instrument of foreign policy).

Foreign Policy, “How Gulf Sovereign Wealth Funds Are Shaping the Middle East” / “instrumental capital,” December 2025.

Bloomberg, analysis of Abu Dhabi’s sovereign funds (Mubadala ~$330bn, >300 deals in five years; MGX; the ~$40bn Aligned Data Centers transaction), December 2025.

Reporting on Saudi Arabia’s Public Investment Fund (~$925bn AUM at end-2025; 2026–2030 strategy; $2tn-by-2030 ambition) and on MGX’s ~$100bn AI orientation (Mubadala / G42, 2024).

The venture-intelligence model

In-Q-Tel, public statements on reaching its 800th investment (2025); CIA charter (1999); non-profit structure and congressional funding.

Northwestern Journal of International Law & Business, comment on In-Q-Tel (analysis of the CIA venture-capital model and its ethical and structural concerns).

Reporting on In-Q-Tel’s documented portfolio (e.g., Palantir, MongoDB, Pure Storage, FireEye) and on the model’s ‘anchor customer / beta-tester’ logic.

TechCrunch, on the Defense Innovation Unit (then DIUx) and the Department of Defense’s intention to ‘piggyback’ on In-Q-Tel’s approach (Ash Carter), 2016; and on NATO’s innovation fund and accelerator as a multilateral application of the model.

Regulation as capital — the Brussels Effect

Bradford, A. The Brussels Effect: How the European Union Rules the World. Oxford University Press, 2020; and “The Brussels Effect,” Northwestern University Law Review 107, 2012.

Siegmann, C., & Anderljung, M. “The Brussels Effect and Artificial Intelligence: How EU Regulation Will Impact the Global AI Market,” 2022 (de facto and de jure effects; the AI Act’s high-risk conformity-assessment requirements; ~5–15% of the EU AI market).

Brookings Institution, “The EU AI Act Will Have Global Impact, but a Limited Brussels Effect,” 2023 (the counter-argument: existing markets, standards bodies, and foreign governments as limiting factors).

Policy Review, “Brussels Effect or Experimentalism? The EU AI Act and Global Standard-Setting” (AI-market fragmentation and product divisibility as conditions weakening the effect for AI relative to GDPR).

General reference: the EU General Data Protection Regulation (GDPR) and the EU AI Act as instances of market-access-driven global standard-setting. The reframing of the Brussels Effect as a species of ‘capital’ is the author’s original analytical contribution.