Google built the world’s most popular search engine. It dominates email, maps, mobile operating systems, and online video. But in the race to put a generative AI assistant on every smartphone, it’s falling behind — and not just to OpenAI.
Gemini, Google’s flagship AI chatbot app, has slipped to third place in U.S. downloads among AI assistant applications, trailing both OpenAI’s ChatGPT and Elon Musk’s Grok. The decline isn’t marginal. It’s structural. And it raises uncomfortable questions about whether the company that arguably pioneered modern AI research can actually win the consumer product fight that matters most.
According to data from Appfigures reported by Yahoo Finance, Gemini’s combined U.S. downloads across iOS and Android fell to roughly 4.6 million in the first five months of 2025, down sharply from approximately 8 million during the same period in 2024. That’s a 42% decline year over year — a staggering number for a product backed by one of the most powerful technology companies on Earth.
ChatGPT, meanwhile, pulled in about 51 million downloads in the same window. Grok managed approximately 8.3 million. Even Meta AI, which launched its standalone app only recently, has been gaining traction at a pace that should worry Mountain View.
The Distribution Paradox
Here’s what makes Gemini’s position so puzzling: Google has distribution advantages that most competitors would kill for. Android commands roughly 45% of the U.S. smartphone market and far more globally. Google Search handles billions of queries daily. Gmail, Google Maps, YouTube — the company’s products touch nearly every internet user on the planet.
And yet none of that has translated into dominance for Gemini.
Part of the problem is strategic confusion. Google has spent the last two years toggling between brand names — Bard became Gemini, Gemini got folded into Search, then pulled partially back out — in a way that has left consumers unclear on what exactly Gemini is or why they need it. Compare that to ChatGPT, which arrived with a single, memorable identity and has maintained it ever since. OpenAI didn’t have a distribution empire. It had a great product and a clear pitch.
Grok’s rise is arguably even more telling. Musk’s AI chatbot is tightly integrated into X (formerly Twitter), giving it a built-in audience of power users who are, by definition, heavy content consumers. But X’s U.S. user base is a fraction of Google’s footprint. The fact that Grok has nearly doubled Gemini’s download numbers suggests that product differentiation and cultural relevance are outweighing raw distribution muscle.
Grok has leaned into a personality — irreverent, unfiltered, willing to engage with topics other chatbots avoid. Whether you find that appealing or reckless, it’s distinctive. Gemini, by contrast, often feels like a corporate product designed by committee. Safe. Capable. Forgettable.
The download numbers don’t capture the full picture, of course. Google has been embedding Gemini capabilities directly into its existing apps — Search, Workspace, Android itself — which means millions of users interact with Gemini-powered features without ever downloading a standalone app. Google has made this argument repeatedly, and it’s not wrong. But it’s also a deflection. The standalone app market is where consumers signal intent. Downloading an AI assistant means you want it as a primary tool, not as a background feature buried inside a search results page.
OpenAI understands this. The company has been aggressive about making ChatGPT the default AI companion across devices. Its partnership with Apple to integrate ChatGPT into Siri and iOS represents perhaps the most consequential distribution deal in the AI era so far. That deal effectively gives OpenAI a presence on every iPhone — the same kind of structural advantage Google has long enjoyed with Android, but now working against it in AI.
A Revenue Problem Hiding Inside a Product Problem
The consumer app race isn’t just about prestige. It’s about money.
AI companies are betting that chatbot apps will become the next great subscription business, following the playbook established by Spotify, Netflix, and the cloud storage providers. ChatGPT Plus costs $20 per month. OpenAI reportedly crossed $5 billion in annualized revenue earlier this year, driven substantially by consumer subscriptions. Grok is bundled with X Premium, giving Musk’s xAI a monetization path tied to an existing subscriber base.
Gemini Advanced, Google’s premium tier, is priced at $19.99 per month as part of the Google One AI Premium plan. But Google hasn’t disclosed subscriber numbers, and the download trajectory suggests uptake has been disappointing. Fewer downloads mean fewer users entering the conversion funnel. Fewer users in the funnel mean fewer paying subscribers. The math is unforgiving.
Google’s overall AI revenue picture remains strong, driven primarily by cloud services and enterprise contracts. Alphabet CEO Sundar Pichai has repeatedly emphasized that AI is boosting engagement across Google’s core products, and the company’s advertising business continues to generate enormous cash flow. But the consumer AI assistant market is a distinct battleground, and losing it could have consequences that ripple outward for years.
Consider the dynamics. If ChatGPT becomes the default AI tool for a generation of knowledge workers, students, and creators, those users will build workflows, habits, and preferences around OpenAI’s platform. Switching costs will compound. The same network effects that made Google Search unassailable in the 2000s could make ChatGPT unassailable in the 2030s — but this time, Google would be on the wrong side of the moat.
Recent moves suggest Google recognizes the urgency. At its I/O developer conference in May 2025, the company unveiled a barrage of Gemini updates: improved reasoning capabilities, tighter integration with Android, new agentic features that allow Gemini to take actions on behalf of users across apps. The company also announced Project Astra, its vision for a multimodal AI assistant that can see, hear, and respond in real time through a phone’s camera and microphone.
These are impressive technical demonstrations. But technical capability has never been Google’s problem. Execution at the product level has. The company has a long history of building brilliant technology and then fumbling the consumer packaging — Google+, Google Wave, Allo, Stadia. The list is long and painful.
Gemini doesn’t need to be added to that list. But the download numbers are a warning sign that can’t be dismissed as a measurement artifact.
There’s also the competitive pressure from Meta. Mark Zuckerberg has made AI assistants a centerpiece of Meta’s product strategy, embedding Meta AI across WhatsApp, Instagram, Facebook, and Messenger — platforms with billions of monthly active users. Meta’s standalone AI app, while newer to the market, benefits from the same kind of social distribution that propelled Facebook itself. And unlike Google, Meta doesn’t have to worry about cannibalizing a search advertising business while promoting its AI assistant.
So where does this leave Google? Technically formidable. Strategically muddled. And, for the first time in a generation, playing catch-up in a product category that may define the next era of personal computing.
The company has the resources, the talent, and the infrastructure to compete. What it hasn’t demonstrated — yet — is the product instinct to win. In consumer technology, being the best model on benchmarks doesn’t matter if nobody opens your app.
What Comes Next
The AI assistant market is still young. Download rankings will shift. New entrants will emerge, and existing players will iterate. Nothing is settled.
But momentum matters. And right now, the momentum belongs to OpenAI and, surprisingly, to Grok. Google’s window to establish Gemini as the default AI assistant isn’t closed, but it’s narrowing. Every month that ChatGPT adds millions of users while Gemini loses them is a month that makes the eventual turnaround harder.
Google has been here before — late to social networking, late to messaging, late to cloud computing. It caught up in cloud. It never caught up in social or messaging. The question hanging over Gemini is which of those patterns will repeat.
For a company that has spent decades organizing the world’s information, the most important information right now might be its own download charts. They’re telling a story Google doesn’t want to hear.
WebProNews is an iEntry Publication