When JPMorgan Asset Management reported that artificial intelligence spending fueled two thirds of US GDP growth in the first half of 2025, it signaled how deeply AI has become part of the economy. Companies across every sector are pouring money into AI transformation. Some analysts see hints of bubble-era enthusiasm, especially after OpenAI’s Sam Altman, Amazon’s Jeff Bezos, and Goldman Sachs CEO David Solomon each acknowledged growing market froth. Even so, recognizing an overheated market is not the same as dismissing AI’s value for businesses.
Corporate AI spending reached 252.3 billion dollars in 2024, according to Stanford University, and private investment jumped almost forty five percent. The question for leaders is no longer whether to back AI. It is how to invest without falling into the traps that leave most companies with little to show for their efforts.
Why only 5 percent of companies succeed with AI
An MIT study found that 95 percent of companies investing in AI have not earned meaningful returns. That number often dominates headlines, but the more useful insight is what separates the top performers who do profit.
ABC News
High performing companies commit more of their digital budgets to AI and behave differently from the rest. McKinsey research shows that three quarters of these leaders have already scaled AI across their operations, compared with only one third of other organizations. They aim for transformative change rather than small efficiency gains. They redesign workflows around AI capabilities and set up strong governance frameworks early. In short, they approach AI as a business transformation strategy instead of a technology upgrade.
The infrastructure challenge
Building cutting edge infrastructure is expensive. Google’s Gemini Ultra cost about 191 million dollars to train. OpenAI’s GPT-4 required nearly 78 million dollars in hardware alone. For many enterprises, building custom large language models is unrealistic. That makes vendor selection and strategic partnerships essential.
Capacity shortages add another layer. CoreWeave cut its 2025 capital spending forecast by up to 40 percent because of delays tied to power infrastructure. Oracle confirmed it is still turning away customers due to limited capacity. Companies that diversify their AI infrastructure partners and test alternative architectures will weather these constraints better than those tied to a single provider.
Investing wisely in a hot market
Goldman Sachs analyst Peter Oppenheimer points out that today’s leading AI companies generate real profits, unlike many speculative firms during the early 2000s. Stock prices have climbed, but earnings have kept pace. For enterprise buyers, the takeaway is not to avoid AI. It is to avoid unfocused spending.
Leaders who see strong returns follow a few consistent principles:
Target clear use cases with measurable ROI. High performers are more than three times as likely to use AI to deliver transformative change rather than experimenting without direction.
Invest in readiness, not only the technology. Companies with agile product teams, strong data infrastructure, and clear talent strategies see the biggest gains.
Build governance from the start. More organizations are taking steps to manage privacy, explainability, reputation, and regulatory risks. Early governance is becoming a competitive advantage as regulations tighten.
Learning from market concentration
By late 2025, five companies made up nearly 30 percent of the S&P 500. This level of concentration makes diversification even more important. The companies in the successful five percent spread their AI bets across multiple vendors, combine cloud and edge approaches, and build internal capabilities for the workflows that matter most.
The strategy that works
Google CEO Sundar Pichai captured the current moment well. He noted that the internet saw plenty of excess investment in its early years, yet no one questions its impact now. AI appears headed down a similar path.
OpenAI’s ChatGPT now has roughly 700 million weekly users, one of the fastest adoption curves in tech history. For enterprises, the challenge is using tools like this effectively, not contributing to the billions wasted on poorly defined projects.
The companies winning with AI treat it as a transformation initiative. They set clear goals, track success metrics, manage change across teams, and stay realistic about vendor promises while still believing in AI’s potential.
What this means for enterprise strategy
Debating whether we are in an AI bubble matters less than building sustainable AI capabilities. Markets correct themselves. Companies that build real AI maturity during this investment surge will stay ahead whether prices rise or fall.
Stanford’s data shows AI adoption jumped to 78 percent of organizations in 2024, up from 55 percent the year before. Waiting for perfect conditions risks falling behind competitors already gaining ground. The priority should be practical deployments, measurable outcomes, and steady capability building.
