The AI boom is Raising the stakes for capital allocation

The AI boom is Raising the stakes for capital allocation

Mon, August 31, 2026 at 11:09 PM GMT+3 5 min read

• GS -1.80%
• IT -4.13%

Artificial intelligence has entered a capital-intensive new phase. Goldman Sachs Research estimates global AI investment could reach about $1 trillion in 2026, underscoring the scale of infrastructure and development now underway. As access to more capable models and computers expands, the time and marginal cost of many execution‑focused tasks may continue to fall. Gartner highlights domain‑specific AI, physical AI, and intelligent simulation as emerging trends, potentially signaling a shift toward more specialized and embodied applications. These developments seem to sharpen a strategic question for business leaders. With powerful tools becoming broadly accessible, deciding where to concentrate capital, talent, and organizational focus may increasingly define competitive advantage.

That question may become more consequential in an environment where the conditions surrounding a decision can change quickly. EY has described supply chains as operating amid geopolitical disruption, trade uncertainty, regulatory volatility, cyber threats, and hidden dependencies, while also pointing to broader shifts in tariffs, markets, and global operating conditions. AI may help organizations process information and identify patterns within that complexity, yet standard predictive systems can still face difficulty when the actors influencing an outcome respond to one another. Historical data may offer a useful starting point, although it can provide only a partial guide when competitors revise their positions, regulators introduce new considerations, or boards alter plans in response to the same event.

The stakes may be especially high as capital allocation decisions grow larger and more interconnected. EY's 2026 M&A Activity Report describes renewed momentum in strategic transactions, including activity across technology, power and utilities, life sciences, and aerospace and defense. Acquisitions, divestitures, market entries, and major investment commitments can involve assumptions about how several participants may react after a decision becomes public. A conventional diligence process may document available information thoroughly, yet a static analysis can have limits when the decision itself changes the environment being analyzed. For capital allocators, the challenge may therefore extend beyond identifying the most likely outcome to considering a wider range of possible responses.

This is where simulation seems to attract increasing attention. Gartner has identified intelligent simulation as an emerging area within decision intelligence, particularly for addressing complex problems. Meanwhile, Forbes reported that venture investors committed more than $3 billion during the first half of 2026 to startups developing "world models," systems designed to model environments and how they evolve. Together, these developments may point toward a broader shift in how AI is being applied to complex decisions.

Artur Kiulian, founder and CEO of Principle, an AI-powered strategic simulation platform, sees this shift as increasingly relevant to corporate and investment decisions. "Every dollar a corporation or investor commits is a bet on a direction," he explains. "If everyone can execute anything, execution stops being the moat. The expensive skill becomes direction, and every capital decision is a bet on a direction." From his perspective, the growing availability of AI tools may increase the value placed on judgment about where capital should go, particularly when an organization is choosing among acquisitions, new markets, funding priorities, or risk positions.

Kiulian argues that the limitations of conventional AI become more apparent when a decision depends on dynamic reactions. "A language model is a compression of the past," he says. "Ask it about the future, and it hands you yesterday, said with confidence." His concern is less about whether historical information has value and more about the risks of treating a single forecast as sufficient for a high-stakes decision. "A single, confident number is what can put allocators at risk in the first place, encouraging them to stake real resources on one outcome and hoping it lands," he says.

Kiulian says that Principle's work reflects an alternative framework built around strategic rehearsal. He sees modeling relevant participants, including competitors, regulators, and boards, as actors with distinct incentives and then examining how their decisions might interact across repeated simulations. "We stopped asking the model what would happen and started using it to rehearse how the world responds," he says. "You don't get a singular prediction. You get a distribution of futures and a strategy that can hold up across most of them." In this context, simulation is intended for decisions involving substantial capital and multiple influential participants, which Kiulian distinguishes from consumer or marketing simulations focused on individual preferences.

He points to an example involving an 18-year-old software company whose products serve more than 30 million users worldwide and which was evaluating the implications of a significant industry platform shift. Kiulian says the exercise explored several possible developments, the incentives affecting major participants, and the strategic choices available under different conditions. The value, he says, extended beyond the analysis itself because leaders could engage directly with the scenarios and discuss the implications before committing resources. He says those conversations eventually touched on practical questions involving acquisitions, investment priorities, and legacy expenditures.

The broader implication may be that capital allocation is becoming a more dynamic exercise. As AI expands access to information and accelerates execution, information asymmetry may provide a different kind of advantage than it once did. "The question is less 'what will happen' and more 'where to put resources given how every other actor might move,'" Kiulian says. His view is that the relevant edge may increasingly involve understanding strategic interactions before they fully unfold.

For leaders navigating larger technology investments and more variable operating conditions, that perspective may encourage a wider use of scenario-based thinking. The goal may be less about finding a perfectly confident answer and more about preparing decisions that can remain useful across several plausible futures.

Ultimately, as AI, automation, and simulation continue to develop, the quality of execution may become easier to scale. The harder task may be deciding which direction deserves commitment when markets, institutions, and competitors can all influence the path that follows.