The promise of generative artificial intelligence has put the U.S. at the center of a potential economic windfall—and antitrust policy is emerging as a pivotal swing factor. An analysis cited by the Computer & Communications Industry Association (CCIA) estimates that “AI-related products and improvements will contribute $15.7 trillion to the global economy by 2030, including $3.7 trillion to the U.S. economy (14.5% of total estimated GDP).” Yet, Forbes and other observers warn that overly broad enforcement could deter the very investments and industry arrangements that make those gains possible.

The AI sector is best understood as a layered "stack" of hardware, cloud infrastructure, foundational models and downstream applications. Competition is intense at each level: hyperscalers such as Google, Amazon and Microsoft are pouring capital into datacenters and custom servers; NVIDIA and other chipmakers are racing to supply the specialized accelerators that dominate large-model training; and dozens of startups are building niche models and services that rely on access to compute, data and distribution channels. That interdependence—scale in hardware and compute enabling model development, which in turn powers applications—is central to the business case for large investments.

Antitrust authorities have legitimate tools to prevent anticompetitive mergers and exclusionary conduct, but economists and industry executives warn of unintended consequences if enforcement is calibrated too bluntly. Training state-of-the-art models can require months of GPU time and capital expenditures running into the tens or hundreds of millions of dollars for compute, staff and data licensing. Joint ventures, licensing deals and cloud partnerships often spread those costs and speed deployment. If courts or regulators treat such arrangements as presumptively suspect, the theory goes, firms will hesitate to enter collaborations that are welfare-enhancing.

“Cutting off the channels for scale and cooperation risks imposing a tax on innovation,” said one industry analyst, speaking on condition of anonymity to discuss sensitive commercial strategy. The concern is not purely theoretical: investors and executives already cite regulatory uncertainty as a factor shaping where they build data centers, sign chip supply contracts or place mergers.

Policy choices can balance these risks. A tailored antitrust approach would focus on exclusionary conduct—tying, predatory pricing, denial of essential inputs—rather than penalizing scale or vertical integration per se. Faster, clearer guidance on when data-sharing, interoperability commitments and conditional approvals are permitted could reduce chilling effects. Remedies that promote open access to essential infrastructure, time-limited licensing requirements or firewalls for sensitive business lines can target harms without dismantling productive cooperation.

Longer term, the structure of AI markets will depend on technological change as much as legal rules. Dominance based on proprietary chips or model architectures can erode if rivals innovate cheaper compute, open-source models or new distribution channels. But that contest requires capital and the willingness to partner. Given the CCIA’s projected $3.7 trillion uplift for the U.S., the policy task is consequential: preserve competitive markets while ensuring that antitrust oversight does not become a headwind to the investments that underwrite national economic leadership in AI.