According to ExtremeTech, OpenAI published its first report on the state of enterprise AI last Monday. The data shows that 800 million people now use ChatGPT weekly, based on usage patterns and a survey of 9,000 workers from nearly 100 companies. For most, the touted time savings are modest: about three-quarters of workers report AI saves them between 40 to 60 minutes of work per day. However, power users are slashing over 10 hours a week. Despite these figures, ChatGPT Enterprise use has exploded, with messages sent jumping roughly 8x and individual users sending 30% more messages on average. The report also highlights specific gains, like IT workers solving problems 87% faster and engineers delivering code 73% faster.
The modest math of productivity
So, an hour a day. Is that good? It’s the question everyone’s asking. On one hand, saving an hour per employee daily is a massive aggregate win for a large company. That’s time for more work, or less stress, or maybe just getting out the door on time. But on the other hand, after all the hype about AI revolutionizing work, it feels… incremental. Here’s the thing, though: the real story isn’t in the average. It’s in the extremes. The “heavy users” saving 10+ hours a week are likely those who’ve fully integrated AI into complex workflows—think coding, data analysis, content creation. For them, it’s a game-changer. For the casual user asking for email drafts, it’s a handy tool. That disparity tells us adoption is still shallow for many.
Where AI actually wins
Look at the specific benefits OpenAI chose to highlight. They’re not about saving 55 minutes on busywork. They’re about dramatically accelerating specialized, high-skill tasks. IT solving problems 87% faster? That’s huge for operational efficiency. Engineers coding 73% faster? That potentially changes product roadmaps. And that stat about 75% of users completing tasks they “couldn’t have done before” is quietly the most profound. It suggests AI is acting as a capability multiplier, not just a time-shifter. This is where the real enterprise value gets unlocked, far beyond simple time tracking.
Adoption vs. optimization
Now, the most telling insight from the OpenAI’s 2025 “The state of enterprise AI report” might be the last one. OpenAI says the main constraint is no longer model performance, but “organizational readiness.” Basically, the tech is good enough. The bottleneck is us. Companies don’t know how to implement this stuff effectively. This explains the gap between modest average savings and explosive growth in usage (140% YoY in countries like Australia and Brazil). Everyone’s piling in to experiment, but turning that experimentation into streamlined, company-wide productivity is a whole other challenge. It’s a people and process problem now.
The global picture and what’s next
The global growth numbers are staggering and hint at where the market is heating up. Japan becoming the largest corporate API hub outside the U.S. is a major data point. But what happens next? I think we’ll see the averages creep up as best practices spread and those “power user” workflows become more common. The companies that figure out the “organizational readiness” puzzle will pull ahead. For everyone else, AI might remain a neat tool that saves an hour a day—which, let’s be honest, is still pretty great. But is it revolutionary? Not yet. The potential is clearly there, lying in wait for businesses to finally catch up.
