April 4, 2026
Generative AI is the big tech story of the decade, that much is clear. While AI has existed since the 1960s, we’ve seen several AI winters, where research and usage dropped off. This changed when ChatGPT was first released. Since then, not only has everyone kept talking about AI, but everyone has also been using it.
The reap of real benefits from artificial neural networks started with this release. And even though we don’t even fully understand why ChatGPT works the way it does (which we’ll discuss soon), it does work. But there’s a key point many people overlook:
Despite large investments into AI research, there is still no measurable boost in revenue.
Before we dive into what benefits generative AI brought us, let’s first check how invested we are in generative AI. What does the current state of AI investment look like? Specifically, how much money have companies put into AI, and what returns have they seen? The numbers might surprise you.
Investments are huge. From 2014 to 2021, AI investments jumped from $18 billion to $119 billion. It’s staggering. Even more astonishing, Big Tech plans to spend about $635 billion on data centers, chips, and other AI infrastructure in 2026 alone (source).
But the shocking figure isn’t the spending. It’s that measurable results are almost nonexistent. According to a report by Business Insider, the actual payoff is zero. That’s right, nada.
Does this mean AI is useless? Well, it’s not that simple.
Although the return on investment looks zero on paper, companies still gain from their investments. The dollar returns may be zero, but productivity benefits are significant. But how do you measure productivity? Increase in quality? Increase in volume? How do we know those reports are accurate? Did companies really boost their productivity, leading to $4.4 trillion in unrealized productivity gains?
What does “unrealized productivity gains” mean in simple terms? It means companies notice higher productivity. This includes tasks like coding, reviewing contracts, or content marketing. However, they can’t turn this productivity boost into revenue yet. Essentially, they’re unsure how to use this increase. They remain stuck in lengthy business processes, which may delay profits from AI.
I’m not saying those numbers are false, but that it’s hard to measure any real productivity gain because productivity cannot be reduced to a number. It’s like someone tried to measure the quality of a product that has 10 different ways of measuring its quality. Depending on which way of measuring you choose, the results will be different as well. That’s why those numbers are not very trustworthy, if at all. If someone asked me which number matters more, I’d say it’s the actual return, and that number is zero.
We built a strong foundation. Even without investments from Big Tech companies like Google and Meta, we still managed to develop several well-designed Large Language Models that improve as we speak. OpenAI released ChatGPT-5.4 on March 27th, breaking its own benchmarks once again. Although it may seem like the only increase since 2022 is in productivity, this isn’t necessarily a bad thing - it means we’re getting benefits. We now need to figure out how to use these benefits with greater efficiency. The next step in the world of AI is to turn this productivity boost into real results and better, higher-quality work.
And we should not forget that generative AI has its downsides too. Although we still cannot tell the exact side effects it has on the human brain, if we ask AI for almost all day-to-day problems, it is going to impact the way we think and solve problems. It is a good idea to keep doing creative tasks manually, without the help of AI, even if it decreases our productivity. Sometimes it’s better to slow down, even if time is of the essence.