Most companies are burning cash on AI and seeing nothing in return. Not slow returns. Not delayed returns. Zero returns. That is the finding from Apollo Global Management’s chief economist Torsten Slok, who looked at the data and concluded that for the vast majority of businesses, AI spending is producing no measurable financial benefit. Billions spent. Profits unchanged.
Why This Story Matters Right Now
Corporate AI spending has been one of the biggest financial stories of the past two years. Companies have raced to announce AI initiatives, hire AI teams, and sign massive software contracts. The market rewarded them for it. Stock prices moved. Analysts cheered.
But Slok’s analysis cuts through the noise. His research, published through Apollo’s economic commentary, argues that the productivity gains most executives promised their boards simply have not shown up in the financial statements. According to Goldman Sachs research from 2024, companies have collectively spent over $1 trillion on AI infrastructure with returns that remain difficult to quantify at the macro level.
According to Gartner, fewer than 30% of AI pilot programs ever reach full production deployment. The rest get shelved. That is a lot of expensive experimentation with no payoff on the books.
We are in a moment where the gap between AI hype and AI results is becoming impossible to ignore. The economists are noticing. The CFOs are starting to ask harder questions.
The Contrarian Take Most People Are Missing
Here is what I think is actually happening. The companies spending the most on AI are not getting returns because they are doing it wrong. They are buying tools before they have a problem worth solving. They are hiring AI teams before they have clean data. They are announcing AI strategies before they have any idea what output they want.
This is not an AI problem. This is a management problem wearing an AI costume.
Think about what happened with websites in 1999. Most companies spent fortunes building web presences that generated zero revenue. Then a smaller group of companies figured out what the web was actually good for, and those companies built enormous businesses. The difference was not the technology. It was the clarity of the use case.
The rich mindset versus the poor mindset shows up clearly here. The poor mindset says: everyone else is doing AI, so we need to do AI too. Then money leaves the building and nothing comes back. The rich mindset says: what specific dollar amount am I losing right now because of a process that AI could fix? Then you build toward that exact number.
According to MIT Sloan Management Review research, companies that tied AI projects to a single specific measurable business outcome before starting saw three times higher rates of meaningful ROI compared to companies that deployed AI broadly without a target metric.
Three times. That is not a marginal difference. That is the difference between a project that justifies itself and one that gets killed in the next budget review.
The other thing the data shows is that small and midsize businesses are actually better positioned to get real AI returns than large enterprises. Why? Because they move faster, they have fewer legacy systems to integrate with, and they can make a decision about a tool in a week instead of eighteen months. If you are a small operator who wants to start producing video content with AI to build an audience, something like InVideo AI lets you go from idea to finished video in an afternoon. No six-month procurement process. No cross-functional working group. Just a tool that does the job.
That is the version of AI that pays off. Specific. Fast. Tied to a real output you can measure.
What This Means for You
If you are running a business or building one, the Apollo finding should change how you think about AI spending. Here is what I would do.
First, stop buying AI subscriptions because they seem smart to have. Every tool needs to answer one question: what does this replace or improve, and how will I know if it is working? If you cannot answer that in two sentences, do not buy it.
Second, start with your single biggest time sink. What task eats the most hours per week in your business for the least strategic value? That is where AI ROI lives. Customer support scripting, content production, data formatting, report summarizing. Pick one. Run a 30-day test. Measure the before and after in actual hours or actual dollars.
Third, do not spend on AI tools that require long-term subscriptions until you have proved the use case works. Sites like AppSumo regularly feature AI tools at one-time prices, which means you can test legitimate software without committing to a recurring cost that drains margin before you have proof it works.
Fourth, ignore the AI announcements coming out of large enterprises. The fact that a Fortune 500 company launched an AI initiative tells you nothing about whether that initiative will generate returns. According to the Harvard Business Review, 80% of enterprise AI projects either fail to meet their stated goals or are quietly abandoned within two years. That number should recalibrate how much weight you put on competitor AI announcements.
The companies that will win with AI are not the ones spending the most. They are the ones who stayed patient, picked specific problems, and built a clear feedback loop between what they spent and what they got back.
The Bottom Line
An Apollo economist just said out loud what a lot of CFOs are whispering in private. AI is not printing money for most companies. It is becoming another line item in a budget full of line items that never quite delivered what the sales deck promised. The businesses that figure out how to make AI pay will not be the ones who spent the most. They will be the ones who stayed specific, measured obsessively, and refused to let excitement replace accountability.
Frequently Asked Questions
Why are most AI investments failing to produce financial returns?
Most companies are deploying AI without a specific measurable goal tied to the investment. They buy tools first and look for problems to solve second. According to research from MIT Sloan, companies that started with a defined financial target were three times more likely to see meaningful returns than those that deployed AI broadly.
What did the Apollo economist actually say about AI payoff?
Apollo Global Management’s chief economist Torsten Slok analyzed corporate AI spending and concluded that for the majority of businesses, the spending has produced zero measurable financial benefit. His research argues that the productivity gains promised by AI vendors have not shown up in company financial statements at the macro level.
Does this mean AI is a bad investment for small businesses?
Not at all. The zero payoff finding is concentrated in large enterprises that are deploying AI broadly without a clear use case. Small businesses that pick one specific problem, use a targeted tool, and measure the result are actually better positioned to see fast returns because they move faster and have simpler operations to improve.
How do I know which AI tools are actually worth buying?
Any tool you buy should answer one question before you pay for it: what specific task will this replace or improve, and how will I measure the difference in 30 days? If the answer is vague, the ROI will be vague. Start with free trials and short-term commitments before locking in to anything expensive.
Is the AI spending boom going to collapse?
I think we will see a significant correction in enterprise AI budgets over the next 18 months as CFOs demand proof of returns. That does not mean AI spending goes away. It means the money shifts from broad AI initiatives to specific high-return use cases. The infrastructure spending will continue. The speculative internal AI projects will get cut.
