ChatGPT and Gemini Both Hit a Billion Users, and Gemini Did It on Android's Default Install

Two AI assistants crossed a billion users this month. That headline reads like a finish line. It isn't. The race to reach people is basically done. Now comes the harder part, and the company that got there first is losing the people who built it. Let me catch you up.

The deep cut

• Reaching people is not the same as keeping them. ChatGPT and Gemini both crossed a billion users, so scale is no longer the moat.
• Distribution beats brand. Gemini rode preinstalled Android to a billion while ChatGPT's lead shrank.
• A tender offer is a stall signal. OpenAI's $7B buyback let staff cash out without an IPO.

Two winners, two very different stories

Both assistants hit a billion, but the road there tells you more than the number. Google's Sundar Pichai called Gemini the company's fastest-growing product ever, its 14th to reach a billion. ChatGPT got there first, back in June, but OpenAI buried the news in a bland blog post instead of celebrating it.

The growth curves point in opposite directions. OpenAI had 900 million weekly users in February and took five months to add the next 100 million. That is a real slowdown for a product once sold as the fastest-growing software ever. Gemini went from 750 million to a billion in the same stretch. The simple read: ChatGPT is bigger, Gemini is climbing faster.

The Android tailwind ChatGPT can't buy

Gemini's number leans on a channel OpenAI does not own. Those billion Gemini users are the app itself, and the app ships preinstalled on every Android phone. Gemini chief Josh Woodward said only 100 million of them are on iOS, which means the rest are Android. Google put the assistant where the phone already is.

OpenAI has no phone. That is why it keeps talking about building hardware. When your rival's growth comes from a default install on billions of devices, matching it with a better model gets you only so far. This is the part your team should sit with: distribution can beat quality, and Google controls the shelf.

Racing to be free instead of first

With the land grab over, the fight moved to what people can do without paying. OpenAI is giving free and Go users unlimited text chats next week, dropping the rate limits that used to stop casual users cold. Most ChatGPT users never pay, so the free tier is where habits form.

There's a tradeoff baked in. Free users move to GPT-5.6 Luna, a model built for "cost-sensitive, high-volume workloads." Cheaper to run, less capable, with a "Think" button to push harder when needed. Paid Sol users get fewer errors, 62% fewer factual mistakes for Luna and 68% for Sol versus the old Instant model. The bet: give away volume, keep the quality behind the paywall.

The bench that keeps clearing out

While the product news looks strong, the org chart is thinning. Brad Lightcap, who joined in 2018 and served as CFO then COO, is leaving to "start something new." He built the first versions of Finance, Legal, People, and Partnerships. That is a lot of institutional memory walking out the door.

He is not alone. AGI chief Fidji Simo left in July, CMO Kate Rouch left in April for health reasons, and Sora and Science leads Bill Peebles and Kevin Weil are gone too. The company is cutting "side quests" and reorganizing around revenue as it preps to go public.

The IPO that keeps not happening

The money moves say wait. OpenAI completed a $7 billion employee tender offer at an $852 billion valuation, flat from its March round. A tender lets staff cash out without a public offering, which is a sign the IPO is not close. The company filed confidentially in June, but this buyback buys time.

Sam Altman admitted "we did not have our best 12 months ever, which is mostly my fault." The Wall Street Journal reported OpenAI missed internal financial goals, and rival Anthropic was reportedly profitable earlier this year. You do not go public on a bad year with a profitable competitor waiting.

Three questions for your team

• If a rival can win on a default install like Gemini did on Android, where are we depending on someone else's shelf, and what happens when they change it?
• Our free tier shapes the habit and our paid tier holds the quality. Are we clear on which users we are willing to serve with a cheaper model, and where that line hurts us?
• When a founding operator like Lightcap leaves, what breaks that nobody wrote down? Do we know which parts of our own org live in one person's head?