What Will Endure: The Pillars of AI’s Economic Shift

By Adena Friedman 
Chair and Chief Executive Officer at Nasdaq

Every breakthrough begins with uncertainty. Railroads once seemed impractical, the telephone unnecessary, and the internet speculative. Yet each became infrastructure we can’t imagine living without – and redefined the boundaries of commerce, connectivity and productivity.

Generative AI is at a critical juncture – one that demands optimism but also invites sharp questions. Will today’s breakthroughs translate into lasting productivity gains? Which business models will survive? How much capital will its buildout demand, and can the investments be justified over the long term? These questions matter because every technological breakthrough carries the risk of short-term exuberance – and, if expectations outpace the ultimate reality – the potential for significant wealth destruction.

Not all investment cycles are created equal. Some are built on the illusion of risk-free returns destined to collapse once the excitement fades, such as the mortgage-backed securities boom in the mid 2000’s. By contrast, investment cycles tied to true technological inflection points – like railroads or the internet – leave the world permanently changed. AI belongs to this second category. Conviction in AI’s transformational potential will fuel capital formation and long-term wealth creation but also calls for discipline – to separate what is enduring from what is ephemeral.

The numbers tell the story: AI will require a capital-intensive buildout on par with past industrial revolutions. Its estimated global AI investment surpassed $1.4 trillion in 2025, and hyperscalers are committing hundreds of billions more to secure compute, energy, and talent. Semiconductor leaders have added over $7 trillion in market cap since 2023, while cloud giants are underwriting multi-year infrastructure bets that will lay the foundation for AI’s growth. Public markets are responding in kind, with valuations reflecting the expectation that AI will become transformational, not transient.

As investment in AI accelerates, the question shifts from speed-to-market to staying power: what will endure? History shows that every industrial era is defined not by the novelty that dominates headlines, but by the scaffolding that makes progress resilient and investable. In the AI era, that scaffolding rests on two pillars: first, the capacity of the financial system to provide the capital that this transformative technology demands to produce durable returns; and second, the core drivers of adoption at scale – particularly at the enterprise level.

Capital to Power a Technology Revolution

Recent headlines have fueled worries that financing models across the AI infrastructure landscape signal a bubble; that valuations are driven by a concentrated, self-perpetuating ecosystem. Circular financing agreements certainly deserve a healthy degree of scrutiny, but they don’t indicate a bubble in and of themselves. The AI companies leading the industry today are at the forefront of not only developing the technology that will redefine our economy for decades to come, but they have also been put at the forefront of innovations in other industries, reshaping the energy and industrial infrastructure that will support the economy. This should prompt important questions about how we can improve and accelerate critical infrastructure build-out and expansion. We can’t expect technology companies to continue to drive transformational innovation while also developing their own energy, compute, and data storage infrastructure. Relying upon a broader ecosystem of industry experts in those key areas will become critical to AI technology’s enduring adoption and transformation.

That said, unlike prior transformative investment cycles, the companies that are underwriting these capital-intensive AI infrastructure projects have the strength to do so. Alphabet, Amazon, Microsoft, Meta, and Oracle spent approximately $428 billion on capital expenditures last year – which collectively represented 69% of their annual operating cash flow. They have ample capacity to make major, long-horizon bets on technologies that will drive demand for decades. Their commitment is deep and reflective of a wider durable trend. In the last year, AI-related investment is estimated to have reached nearly 1% of U.S. GDP, but that remains below the levels from past technology buildouts that have generally reached levels of 2-5% of GDP. Further, it’s important to consider that a significant portion of the infrastructure investment – which does not sit on the balance sheet of these technology companies – is led by investors such as Blackstone, Brookfield and sovereign wealth funds with long-term investment horizons as well as strong and steady capital balances.

Financing looks concentrated now, but history tells us it likely won’t stay that way. As the cycle matures, sources of funding will diversify, just as they did for railroads, telephony, and the internet. If you consider the funding environment today – outside of the hyperscalers and semiconductor companies – innovators are relying predominantly on private markets, which offer access to capital that is orders of magnitude smaller than what is available in the public markets. With over $125 trillion of investable dollars, public markets stand to become the primary engine for scaling AI, offering what private capital cannot at global scale: liquidity, transparency and efficiency.

Adoption – and Gains – at Scale

The adoption of generative AI by consumers is already starting to change how we learn, engage and communicate online, but it’s the proliferation at scale within the enterprise that will enable the technology to reach its full potential and transform the client experience. That doesn’t mean replacing existing software overnight; it means embedding AI into mission-critical processes in ways that are secure, compliant, and explainable. Achieving this requires more than technology – it demands organizational readiness and a regulatory environment that enables projects to move at speed without compromising safety or accountability. Streamlined approval processes, clear standards for data governance, and harmonized compliance frameworks will be critical to unlock scale. When these foundations are in place, adoption moves beyond experimentation to enterprise-wide deployment, creating the conditions for lasting productivity gains.

AI has already crossed the adoption threshold in corporate America. In 2025, nearly nine in ten companies reported regular use of AI, up from 78% a year earlier and the early returns are compelling: three in four enterprises already see positive ROI, with every dollar spent on generative AI delivering nearly 2.8x back. So, the question isn’t whether AI works – it’s how to move from pilots and proof-points to enterprise-wide rollouts.

One crucial hurdle is earning trust and implementing solutions with enterprise-grade security. Only one third of firms have begun to scale their AI programs, and just 7% have fully deployed and integrated it across their organizations. About 80% admit they’re not fully prepared for emerging AI compliance requirements, and 93% lack confidence in securing the data and outputs AI generates. The next chapter isn’t about proving AI works – it’s about making it trustworthy at scale.

In enterprise software, trust is the moat that will turn pilots into scale. Startups will keep pushing the frontier, but platforms that embed compliance, security, and resilience will form the backbone of this transformation. Real impact will require a more transformational view of the enterprise which requires rethinking workflows, governance, and product strategy to maximize value.

An Opportunity – and Imperative – for Transformative Change

An important principle underpinning AI’s widespread optimism is its potential to improve productivity. Paul Krugman once observed, “Productivity isn’t everything, but in the long run it is almost everything,” and he’s right; rising living standards ultimately depend on our ability to grow productivity over time. Yet, global economies have faced persistent productivity stagnation. Despite decades of technological innovation and investment, productivity growth has slowed significantly in most modern economies. This is the backdrop against which AI’s promise should be judged. The promise of AI will be measured in whether we can translate innovation into sustained productivity gains.

Achieving that outcome will require us to think differently. Simply layering AI into existing workflows or taking a plug-and-play approach won’t deliver transformational impact. Like transformational technologies that preceded it, AI’s full potential demands a more fundamental redesign of how work is structured and how value is created.

The story of AI will be written in decades, not quarters, and its trajectory will not be immune to market dynamics. As we have seen with every major investment cycle, we will see winners and losers, but AI will deliver enduring impact. The winners in this era will be those who build trust into enterprise adoption, deploy capital with a long‑term horizon, and rethink their technology, organizational structures, and data architecture from the ground up to unlock AI’s full potential. If history is a guide, the scaffolding being built today will become tomorrow’s indispensable infrastructure. AI’s investment cycle is not a sprint; it is a marathon – and the race has only just begun.