Three years after launching on November 30, 2022, ChatGPT has hit a wild milestone: about 800 million people use it every week. That is the kind of number you would expect to read about in a history recap, not a product update.
On the surface, it looks like ChatGPT is the undisputed king of AI. But the view from the throne is complicated. Google is coming hard with its own AI models, critics keep asking whether this technology is actually reliable, and the business side is burning through billions of dollars.
So yes, ChatGPT sits on top of the AI world, but the ground underneath it is shaking.
An Unprecedented Rise to Power
ChatGPT did not just grow fast. It grew at record speed. It became the fastest-growing consumer app in history.
Younger people drove a lot of that growth. Nearly 60% of adults under 30 have used ChatGPT. When more than half of a generation that lives online uses something, it spreads everywhere: social media, school, work, side hustles, and group chats.
From there, it went mainstream. According to the Pew Research Center, about one-third of all American adults have used ChatGPT. That means it jumped from being a niche tech toy to something your coworkers, friends, and maybe even your parents have tried.
Experts noticed the shift. Anton Dahbura, co-director of the Johns Hopkins Institute for Assured Autonomy, said that ChatGPT helped start a “breakthrough period” for AI. In simple terms, there was life before ChatGPT and life after, and the line between those two is very clear.
The Balance of Power Tilts
While OpenAI celebrates that huge user base, the competitive landscape is changing fast.
Google has become a serious challenger. Its new model, Gemini 3 Pro, launched on November 18 and immediately shook things up by beating GPT-5.1 (the model behind ChatGPT) on important AI benchmark tests.
Think of benchmarks as standardized exams for AI. They test logic, math, reading, coding, and similar skills. On those tests, here is how things looked:
- Gemini 3 Pro score: 1324
- GPT-5.1 score: 1220
If GPT-5.1 used to be the top student in class, Gemini 3 Pro just turned in an exam with a clearly higher score.
This is already having real-life impact.
The CEO of Salesforce, a major figure in the tech world who had been using ChatGPT daily for three years, publicly said he switched to Gemini 3. His reason was simple: he felt it was better at reasoning and faster.
Moves like that send a clear signal. Even long-time power users are willing to change tools if they believe another AI gives them better results.
Inside OpenAI, the pressure is obvious. CEO Sam Altman has reportedly told employees to prepare for “challenging times” and “temporary economic obstacles.” In plain English, he is saying that competition is intense and the next stretch will not be easy.
Cracks in the Code
Despite its massive popularity, ChatGPT still has a serious problem: it is not always reliable.
Critics say that large language models like GPT-5.1 are basically very advanced pattern-matchers. They do not truly understand the world. Instead, they guess what word should come next based on patterns in the data they were trained on. It can look incredibly smart and fluent, but it can also confidently make things up.
A big example is fake citations.
A study from Deakin University found that ChatGPT fabricated about 1 in 5 academic citations it produced. Even more worrying, 56% of all citations it generated were either completely fake or contained major errors.
Imagine asking someone you trust to give you sources for a research paper, and more than half of those sources are wrong or invented. That is not a small glitch. That is a trust issue.
OpenAI’s own leadership has been surprisingly honest about the limitations. In August 2025, Sam Altman said that GPT-5 does not meet his standard for Artificial General Intelligence (AGI).
AGI is the idea of an AI that can learn, reason, and adapt across many different tasks the way a human can, not just produce good text. Altman’s point was that GPT-5 still cannot truly learn on its own. It is powerful and impressive, but it is not “human-level intelligence,” and it still has deep technical limits.
The Billion-Dollar Balance Sheet
Now for the money side, which is just as intense.
Despite all the hype and usage, OpenAI is losing enormous amounts of money.
In just the first half of 2025:
- Net loss: about 13.5 billion dollars
- Revenue: about 4.3 billion dollars
So the company brought in a lot of money, but spent far more than it earned. And this is not expected to turn around quickly. OpenAI’s own financial projections say they expect to keep losing money every year through 2028, with operating losses reaching around 74 billion dollars in 2028. Only after that are they aiming to become profitable, around 2030.
In simple terms, they are making a gigantic, long-term bet that all this spending on AI will someday pay off in a big way. For now, running these models is incredibly expensive.
This pattern is not unique to OpenAI. It reflects a broader issue across the AI industry as a whole.
Research from the MIT Media Lab found that 95% of organizations using AI are not seeing a measurable return on their investment. Many companies believe AI will eventually boost productivity or cut costs, but right now, most are not seeing solid, trackable proof.
There is another hidden cost too. Studies suggest that employees spend about 41% of their time just checking and fixing AI generated content.
So even when AI drafts emails, reports, or summaries, humans spend a lot of time reviewing and correcting that work. The promise of “instant productivity” often turns into “please double-check everything.”
What Comes Next for the AI Leader?
That brings us to ChatGPT’s current position. It sits at a very interesting and very uncomfortable crossroads.
On one side:
- It has an enormous user base, around 800 million weekly users.
- It helped kick off the modern generative AI boom and changed how people think about interacting with computers.
On the other side:
- Google’s Gemini 3 Pro is now scoring higher than GPT-5.1 on key tests and winning over high profile users.
- Its technical reliability is still questionable, especially for serious tasks like research, legal work, and business analysis.
- Its financial path is rough, with huge losses projected for years.
So we are left with a big, open question:
Is ChatGPT simply going through intense growing pains,
or are we watching the start of a major shift in who leads the AI race?
Is it going to remain the main character in the AI story, or become one powerful tool among many in a crowded field?
Whatever happens next, one thing is clear.
The system that helped kick off the current AI revolution is now fighting hard to secure its future in the very world it helped create.