Is ChatGPT Still the Runaway Leader? Why Google Struggles to Catch Up in the AI Race

Is ChatGPT Still the Runaway Leader?

Why Google Can’t Close the Gap in the AI Arms Race

Introduction: ChatGPT Became “the Face of AI” in Just Three Years

Let’s sync our sense of time. The conversational AI “ChatGPT” was opened to the public only three years ago. Today it reportedly has around 800 million weekly active users worldwide and more than six billion monthly visits, making it the most-used generative AI tool on the planet.

Meanwhile, Google—once synonymous with search—has rapidly grown usage for its chat AI, Gemini. So is the title right: “Is ChatGPT still running away with it?” Or will Google steadily close in and eventually overtake it?

In this article we’ll walk through:

1. The current state in numbers (users, share, loyalty)
2. Product design and business-model differences
3. Leadership styles of Sam Altman and Demis Hassabis
4. What “runaway lead” means from a Japanese market perspective
5. Plausible scenarios from here

---

Chapter 1. What’s Happening Now: The AI Race in Numbers

1-1. A Quick Look at Scale: ChatGPT vs. Gemini

First, the rough scale.

• ChatGPT
• Weekly active users: ~800 million
• Monthly visits: ~6 billion (as of fall 2025)
• Around 70% of all generative-AI site traffic

• Google Gemini
• Monthly users: an estimated 600–700 million
• Monthly visits: a little over 1 billion
• Roughly 10%+ of generative-AI traffic, doubling in a year

“Those numbers sound closer than I expected,” you might think. But what matters isn’t just “how many people show up”—it’s “how long they stay” and “how deeply they use it.”

1-2. The Loyalty Gap: 82% vs. 49%

Analyses such as Similarweb show a big loyalty difference among generative-AI sites.

• Roughly 80%+ of ChatGPT users barely touch other generative-AI services.
• About half of Gemini users also use other tools like ChatGPT or Claude.

In other words, ChatGPT is structurally positioned to become “the main AI” or “default AI,” much like many of us feel “ChatGPT alone is enough.”

Think of smartphone OS behavior:

• iPhone users often live almost entirely inside iOS.
• Android users tend to mix and match across manufacturers and apps.

Likewise, ChatGPT is turning into the “home base you open to do everything,” while Gemini can look like “one strong feature inside Google’s huge suite of services.”

---

Chapter 2. The Business-Model Bind That Makes Catching Up Hard for Google

2-1. A Golden Castle of Search and Ads

Google earns most of its revenue from search ads and YouTube ads, with search advertising as the long-standing profit engine.

But what happens if generative AI answers everything on the spot?

• Before: Users click several search-result links → ads show on those pages → Google earns revenue.
• With AI answers: The AI summarizes everything at the top → fewer site clicks → the ad-revenue structure wobbles.

So Google faces a dilemma:

“If we improve the AI experience too much, we undermine our own business model.”

That’s why we see cautious product choices—AI answers in side panels, retaining the classic search UI—designed to avoid a sudden switch.

2-2. ChatGPT Has Little to “Break”

OpenAI, in contrast:

• Has no giant search business to protect.
• Has far more upside in creating new businesses than risk of destroying old ones.

It’s the classic “innovator’s dilemma” Clayton Christensen described:

• Google: a gigantic cash cow that earns money even if left alone.
• OpenAI: desperately cultivating the cash cow itself right now.

Big incumbents have strong incentives to defend the status quo, which makes it harder to go all-in on disruptive tech. The gap between ChatGPT and Gemini reflects exactly this structure.

---

Chapter 3. Cultural Differences: Sam Altman vs. Demis Hassabis

3-1. Sam Altman: The Startup Growth Architect

OpenAI CEO Sam Altman, former president of Y Combinator, brought a culture distilled to:

“Ship fast, learn from users, rebuild quickly.”

ChatGPT embodies that playbook:

• Ship even beta-like features to the public.
• Absorb feedback—criticism and blowback included—and improve on a cadence of weeks to months.
• Polish what works at speed, discard what doesn’t.

ChatGPT is effectively “the world’s largest experiment lab,” and users are invited in as co-conspirators in its evolution.

3-2. Demis Hassabis: A Researcher Who Leads With Science

Demis Hassabis, who heads DeepMind under Google, is a researcher known for breakthroughs like game AIs and protein-structure prediction with AlphaFold.

His comments often emphasize:

• Treating AI as a technology advancing at “ten times the scale and speed of the Industrial Revolution.”
• Creating big scientific breakthroughs through collaboration between human experts and AI.

It’s a mindset centered on the long-term benefit of science and humanity.

So the culture around Hassabis is:

“Build something scientifically solid first, then layer services on top carefully.”

A research-lab approach.

3-3. “Culture Eats Strategy for Breakfast”

Peter Drucker’s line—“culture eats strategy for breakfast”—applies here. No matter how grand the strategy, it fails if the culture doesn’t fit.

• OpenAI: startup culture × continuous global experiments with users.
• Google/DeepMind: research culture × careful product rollouts.

The AI race is not just about model quality; it’s about which culture can improve user experience faster.

---

Chapter 4. Engagement Design: Why ChatGPT Became a “Conversational Partner”

4-1. Designing AI as a Companion, Not Just a Service

ChatGPT’s strength isn’t just intelligence. Key is how it becomes a daily habit:

• You can talk to it anytime.
• Conversation history sits there so you can easily continue.
• You can customize it with your own settings and GPTs.

Those features make “talking with AI” feel like using an everyday messaging app.

Think of your phone’s home screen: you use the apps your fingers instinctively reach for, not the ones with the highest specs.

ChatGPT achieved two things:

1. Strong branding as a standalone site and app.
2. Creating a global habit of “just open ChatGPT first.”

4-2. Gemini Often Looks Like “AI as a Feature”

Gemini’s capabilities are improving fast. It’s integrated into Google Search, Gmail, Docs, YouTube, and more, which is powerful because “you can reach AI from anywhere.”

But here’s the paradox:

• The more it’s embedded as a convenient feature,
• The more complex it feels to remember which service and which button to press.

So we end up with:

• ChatGPT: a single “information desk” where you can ask anything.
• Gemini: present in many places, but its role looks scattered.

Picture a shopping mall where every store has helpful staff, but there’s just one information counter. ChatGPT is the counter itself; Gemini can feel like each store’s staff.

4-3. Measuring “Stickiness”: Loyalty as a Metric

Marketing teaches us it costs more to acquire a new user than to retain an existing one. Loyalty—the difficulty of leaving—is a lifeline.

ChatGPT remains top in traffic share, and over 80% of its users don’t use other services, meaning it secured an “AI search-engine position” very early.

Google built the habit of “just Google it.” OpenAI is recreating that as “just ask ChatGPT,” revealing the irony of this new race.

---

Chapter 5. The Money: OpenAI’s Huge Bet vs. Google’s Cushion

5-1. A Plan to Burn $115 Billion

Behind the AI race lies staggering money. Reports say OpenAI plans to invest about $115 billion cumulatively by 2029, burning over $8 billion in 2025 alone on cloud costs, data centers, and custom chips—nearly unprecedented in tech history.

At that scale you can’t simply slow down because user numbers dip. You must keep growing users and revenue, turning the business into a shark that dies if it stops swimming.

5-2. Already a Massive Business on Revenue

On revenue, ChatGPT is already huge:

• Estimated annual revenue above $10 billion.
• Some expect a $20 billion run rate in 2025.
• Tens of millions of paying users for plans like Plus.

So ChatGPT is simultaneously a massive revenue source and the center of a massive cost burn. That’s why Altman keeps shipping features that pair engagement with monetization—personas, shopping functions, richer paid tiers, and more.

5-3. Google Can Move at the Pace of a Super Tanker

Alphabet, on the other hand, already has multiple revenue pillars—ads, Google Cloud, Workspace, YouTube, and more. The company won’t shake even if Gemini’s growth is slower.

Think of it like this:

• OpenAI: a high-performance speedboat that risks sinking if it ever slows.
• Google: a supertanker that turns slowly but shrugs off waves.

That dynamic explains why ChatGPT can ship features fast while Google stays cautious.

---

Chapter 6. Performance Battles Are Heading Toward a Draw

6-1. ChatGPT 5 vs. Gemini 2.5 on Japanese National Exams

A 2025 paper using Japanese dental national-exam questions compared several generative AIs, including ChatGPT 5 and Gemini 2.5.

Results:

• Both scored above 80% and cleared the passing line.
• Scores improved sharply versus older versions that fell short in 2024.

So at least for specialized exams, both are already “smart enough.”

6-2. Which Means Experience and Ecosystems Decide the Winner

Innovation theory notes that once performance exceeds what users need, competition shifts from performance to convenience, experience, and ecosystem—just as smartphones moved from CPU specs to camera usability and service integration.

In generative AI, benchmark score gaps are a few points; experience gaps—habit formation, stickiness, the sense of a “partner”—may matter far more over time.

ChatGPT looks like it’s running away precisely because it nailed experience design and ecosystem first.

---

Chapter 7. What Does ChatGPT’s “Runaway Lead” Look Like from Japan?

7-1. Japan Is Still an Emerging Market

Japanese surveys show:

• Awareness of generative AI: a little over 70%.
• People who have actually used it: a little over 40%.
• Daily users: still a minority.

So many know about it, but few use it deeply. Market-size estimates go from about ¥118.8 billion in 2023 to roughly ¥1.8 trillion in 2030 (CAGR ~47%). In Japan both ChatGPT and Gemini are still in the early expansion phase.

7-2. Overseas Players Dominate

Government materials note that most generative AI used in Japan comes from overseas providers like OpenAI and Google. Domestic players are growing, but the ChatGPT-vs.-Gemini framing still shapes most first experiences for Japanese users.

In schools and companies, choices like “let’s start with ChatGPT” or “Gemini fits our Google-account policy better” could set future norms and habits.

---

Chapter 8. So Is It Really ChatGPT’s Runaway Yet?

8-1. For Now, ChatGPT Leads in Usage and Loyalty

Across traffic share, weekly active users, and loyalty (the share who use no other services), ChatGPT still holds a near-runaway position.

The brand equation “AI = ChatGPT,” the habit of “ask ChatGPT when in doubt,” and the ecosystem of custom GPTs and plugins are becoming tough assets for latecomers to match.

8-2. Yet the Market Is Gradually Splintering

Still, note this:

Even with a “runaway,” ChatGPT’s share is slowly declining.

Its traffic share fell from over 80% to around 70%, while Gemini grew from single digits to the low teens. New rivals like Claude, Perplexity, and DeepSeek keep emerging. We’re moving from “ChatGPT alone” toward “ChatGPT-led multipolarity.”

8-3. What Would Google Need to Pull Ahead?

For Google to truly catch up—or overtake—it needs three things:

1. Redesign the business model
• Boldly pivot to AI-first search without fearing conflict with search ads.
2. Reinvent engagement
• Elevate Gemini from “a feature” to “a companion.”
3. Tune development culture
• Keep research rigor while embracing user-side experimentation speed.

If those three line up, Gemini could finally shake ChatGPT’s lead.

---

Takeaways: Three Points to Remember

Here are three points to keep after reading:

1. Performance is nearing a draw; the real gaps are engagement and culture.
• ChatGPT 5 and Gemini 2.5 clear many benchmarks.
• The core battle is how much they can be used every day.

1. ChatGPT’s strength is becoming an independent companion woven into daily life worldwide.
• It’s the default AI people open first.
• Conversation history, custom GPTs, and plugins make it sticky; over 80% of users don’t multi-home.

1. Google’s weakness is the flip side of its strengths: huge existing businesses and a research-lab mindset.
• It must protect the search-ad cash cow.
• DeepMind’s scientific culture favors caution over startup-speed experiments with users.

So the current answer to “Is ChatGPT still the runaway leader?” is:

Yes for now in usage and loyalty, but the lead isn’t permanent—the market is already becoming multipolar.

When you pick an AI for work or study, consider:

• Which one will you choose as your main partner (ChatGPT, Gemini, or something else)?
• How much should you depend on a specific service?
• How might regulation and competition shift in Japan?

Thinking through those questions is how each of us builds a small personal strategy for an era of AI power struggles.