When JPMorgan Asset Management declared that AI’s business spending accounted for two-thirds of US GDP growth in the first half of 2025, it was more than a simple statistic—it was a sign.
The discussion began to change recently when OpenAI CEO Sam Altman, Amazon’s Jeff Bezos, and Goldman Sachs CEO David Solomon each recognised market froth within a few days of one another. But here’s the takeaway for enterprise decision-makers: recognising that markets have gotten overheated isn’t tantamount to discounting AI’s business value.
Corporate investment in AI’s business will grow 44.5% to US$252.3 billion globally in 2024, according to Stanford University. The question is no longer whether you should invest in AI’s business, it’s how to invest smartly when others – namely, an enterprise’s competition – overspend on platforms and solutions that may not even ever yield a positive return.
What sets AI’s business success shots apart from the 95% that have failed
According to ABC News, a study from MIT says that 95% of businesses that that have invested in AI will not see a positive return on the technology. But that number hides a more significant truth: 5% win – and they’re playing the game in an entirely different way.
High-performance organisations are spending more on AI capabilities, with over one-third investing 20% or more of their digital budget in AI technologies, according to McKinsey. They’re not just spending more — they’re spending more smartly.
The McKinsey research lays out what winners do differently from the pack. Three-quarters of high performers say their organisations are scaling or have already scaled AI, compared with only a third of all other organisations. The leaders have something in common: they strive for disruptive innovation over incremental improvements, redesign workflows around AI capabilities, and put in place strict governance frameworks.
The infrastructure investment dilemma
Business leaders are confronted with a real puzzle. Google’s Gemini Ultra cost $191 million to produce, and OpenAI’s GPT-4 incurred $78 million in hardware costs alone. Most companies can’t build their own proprietary extra-large language models, which is why vendor selection and partner strategy are so crucial.
Even with soaring demand, CoreWeave cut its 2025 capital spending forecast by up to 40%, citing delays in the delivery of power infrastructure. Oracle is “still having to turn customers away” because it can’t meet demand, CEO Safra Catz said, according to a report by Euronews.
This creates risk and opportunity. Companies that hedge AI infrastructure bets — forging relationships with multiple providers, validating alternative architectures, and stress-testing for supply disruptions — might find themselves in a stronger position than those going all-in on any one hyperscaler.
Frothy market for AI investments that are strategic
Goldman Sachs equity analyst Peter Oppenheimer notes that “unlike the speculative companies of the early 2000s, today’s AI giants are delivering real profits. But even as the stock prices of A.I.-fueled companies have skyrocketed, so has long-term earnings growth.”
What the enterprise takeaway isn’t is to not invest in AI – it’s how not to make the same mistake as 95% of those that see no return:
Concentrate on Use-Case Focus with Measurable ROI:
High-flyers are also over three times as likely as others to report that their company intends to use AI to produce a step-change transformation in their businesses, data from McKinsey reveals. They aren’t using AI for the sake of it; instead, they’re focusing on specific business challenges where AI delivers clear value to their organisation.
Invest in the organisational dimension, not just technology: Having an adaptive product-delivery organisation is strongly associated with value. Developing strong talent strategies and deploying technology and data infrastructure are meaningful for AI success.
Develop governance frameworks today:
The proportion of respondents who say they have implemented mitigation efforts for risks such as personal privacy, explainability, organisational reputation, and regulatory compliance has increased since 2022. As global regulations grow stricter, early governance investment is a competitive edge.
Learning from market concentration
By late 2025, only five companies accounted for 30% of the US S&P 500 index – the highest level in half a century. For the enterprise, this is a dependency that would undoubtedly be worth managing.
The 5% that succeed do so precisely by diversifying their AI vendors and strategic approaches. They’re mixing and matching cloud-based AI services with edge computing, working with several model vendors in partnership, and developing their own expertise for the most critical strategic workflows.
The real AI investment strategy
Google’s Sundar Pichai captured the nuance companies must navigate: “We can gain perspective from looking back at the internet today. There was certainly a lot of overinvestment, but I doubt any of us would say the internet was superficial. I expect AI to be the same.”
OpenAI’s ChatGPT is used by about 700 million people every week and has become one of the fastest-growing consumer products ever. The challenge for the enterprise is how to deploy it effectively, not watching others waste billions on vanity projects.
The one thing AI winners have in common is that Best-in-class companies embrace treating AI as a business transformation initiative, not merely a technology project. Before they roll it out, they define clear success measures. They’re investing in change management as well as infrastructure. And they are rightfully sceptical of vendor promises and committed to potential.
What it means for the business sector
Whether we’re in an AI bubble is less critical to enterprise leaders than ensuring their AI capabilities are sustainable. The market will sort it out – it always has. But the businesses that develop real AI capabilities during this spending surge will be stronger no matter what happens in the broader marketplace.
The share of survey respondents saying their organisations were using AI increased to 78% in 2024 from 55% in 2023, according to the Stanford data. AI is on the move, and enterprises waiting for ideal market conditions will fall behind competitors who start building the needed AI capabilities today.
The strategy is to ensure your AI ROI delivers measurable business value, regardless of market sentiment. Dwell on tangible implementations, quantifiable results and organisational preparedness. Let everyone else chase the valuation bubble while you calmly build a real competitive advantage.
