ChatGPT's 'Code Red' Moment: Three Years After Disrupting Google, the Tables Have Turned

Three years ago this week, ChatGPT broke the internet. Within five days of its November 2022 launch, the chatbot had a million users. Within two months, it had 100 million. Google's CEO Sundar Pichai reportedly declared a "code red," pulling engineers off other projects to respond to what he saw as an existential threat to search.

Fast forward to December 2025, and there's another code red. But this time, it's Sam Altman sounding the alarm.

On 2 December 2025, OpenAI's CEO sent an internal memo telling staff they're "at a critical time for ChatGPT." The company's delaying advertising products, shopping agents, and a personalised assistant called Pulse. Everything's being redirected to making ChatGPT faster, more reliable, and more personal. (CNBC, December 2025)

The irony isn't lost on anyone. The disruptor has become the disrupted.

What Changed in Three Years

Here's the short version: while OpenAI was busy becoming a household name, its competitors were quietly eating its lunch in the places that actually matter.

Anthropic, the safety-focused startup founded by former OpenAI researchers, now commands 32% of the enterprise AI market by model usage. OpenAI? Just 25%. Google sits at 20%. (Menlo Ventures Survey, 2025)

That's a stunning reversal. In 2023, OpenAI held roughly 50% of enterprise market share. They've been cut in half.

The numbers tell an even starker story when you look at growth rates. According to Sensor Tower data, ChatGPT's user growth slowed to just 6% between August and November 2025. Gemini? It grew 30% in the same period. Daily engagement with ChatGPT dropped 10% since July, while time spent on Gemini has more than doubled since March. (Washington Post, December 2025)

The Enterprise Exodus

Why are businesses choosing Claude over ChatGPT? It comes down to reliability.

About 80% of Anthropic's revenue comes from enterprise customers. Their Claude models have become the go-to for coding, document drafting, and workflow automation. Claude Code, Anthropic's developer tool, hit $1 billion in run-rate revenue in just six months. That's not a typo. (Fortune, December 2025)

The company's revenue trajectory is almost absurd. Zero to $100 million in 2023. $100 million to $1 billion in 2024. Projected $8-10 billion by the end of 2025. That's 10x annual growth for three consecutive years.

Now they're preparing for what could be one of the largest tech IPOs ever, with a potential valuation exceeding $300 billion. They've hired Wilson Sonsini Goodrich & Rosati, the same law firm that handled Google's, LinkedIn's, and Lyft's public offerings. (Financial Times, December 2025)

When asked about OpenAI's code red, Anthropic CEO Dario Amodei didn't hold back.

"We don't have to do any code reds," he said. (Analytics India Magazine, December 2025)

Google's Comeback

Then there's Google. Remember when everyone said they'd missed the AI boat?

Gemini 3, released in November 2025, has been turning heads. The "Deep Think" reasoning mode rolled out to Ultra subscribers on 4 December, and early reviews are strong. Andrej Karpathy, the respected AI researcher who's worked at both Tesla and OpenAI, called it "very solid daily driver potential, clearly a tier 1 LLM."

But it's Geoffrey Hinton's comments that really sting. The "Godfather of AI," a Nobel Prize winner who helped invent the deep learning techniques that make all of this possible, now thinks Google will win the AI race.

"My guess is Google will win," Hinton told Business Insider. "Right now they're beginning to overtake it [OpenAI]. Google invented transformers. Google had big chatbots before other people." (Yahoo Finance, December 2025)

Google's advantages are structural. They've got the data. They've got the custom chips (TPUs). They've got 650 million monthly active users on Gemini and climbing. Forrester analyst Mike Gualtieri put it bluntly: "I don't see how OpenAI and Anthropic can overcome that."

The Marc Benioff Moment

Sometimes a single tweet captures a shift better than any analyst report.

On 23 November 2025, Salesforce CEO Marc Benioff posted to X:

The post got 3.2 million views. (Fortune, November 2025)

When a billionaire tech CEO who's been a loyal ChatGPT user publicly defects to the competition, that's not just a product review. That's a signal.

The Capital War

CNBC's Jim Cramer, whose investment advice has famously spawned an entire "inverse Cramer" investment strategy, offered a different take on what's really happening.

"The real code red is a capital war," he said. "All of OpenAI's competitors have better access to credit." (CNBC, December 2025)

He's got a point. (Mark your calendars.) Google has virtually unlimited resources. Anthropic just secured a $15 billion combined commitment from Microsoft and NVIDIA. They're building $50 billion worth of AI infrastructure, starting with data centres in Texas and New York.

OpenAI isn't poor by any means. They've secured over $1.4 trillion in long-term infrastructure commitments. But the gap between "well-funded startup" and "backed by the world's largest tech companies" is showing.

The Netscape Problem

Some observers are drawing a darker parallel. Jason Calacanis, co-host of the All-In podcast, invoked a cautionary tale from tech history:

The Netscape comparison is brutal but apt. In the 1990s, Netscape Navigator owned the browser market until Microsoft bundled Internet Explorer free with Windows. Netscape's 90% market share evaporated within three years.

OpenAI faces a similar dynamic. Google's giving away Gemini. Apple's building AI into every iPhone. Microsoft's integrating Copilot everywhere. Meta's open-sourcing Llama. When your competitors can afford to make their products free, being "slightly better" isn't enough. You need to be dramatically better.

The question is whether ChatGPT can maintain a 3x quality advantage over free alternatives. Right now, most users would say the gap is narrowing, not widening.

The Skeptic's View

Not everyone's buying the narrative.

Eric Newcomer, the tech journalist behind Platformer, has a more cynical read. He suggests the "code red" might be strategic PR, designed to set up a victory narrative when OpenAI launches GPT-5.2, which Altman has promised by 9 December.

It's a fair point. Altman's a master of the media cycle. Declaring a crisis, then triumphantly solving it, is a classic playbook move.

But even if the framing is strategic, the underlying competitive pressure is real. You don't pause your advertising and shopping initiatives because everything's going great.

What This Actually Means

Let's cut through the noise. Here's what businesses should take from this:

The market's fragmenting. There's no longer a clear "best" AI model. GPT-5.1, Claude Opus 4.5, and Gemini 3 are all genuinely competitive. The right choice depends on your use case.

Enterprise and consumer are different games. OpenAI dominates consumer mindshare. Most people still think "AI chatbot" means ChatGPT. But in corporate environments, where reliability and safety matter more than buzz, Anthropic's winning.

Switching costs are lower than you think. Marc Benioff used ChatGPT daily for three years and switched after two hours with Gemini. If a power user can defect that easily, so can your employees. So can your customers.

Multi-model strategies are becoming standard. Smart companies aren't betting on a single provider anymore. They're building architectures that can swap between models as capabilities and pricing shift.

The Three-Year Cycle

There's something almost poetic about the timing. In December 2022, Google declared its code red over ChatGPT. On 2 December 2025, almost exactly three years later, OpenAI declared its own.

In tech, three years is both an eternity and the blink of an eye. It's long enough for a startup to become a household name. It's short enough that the household name can find itself playing catch-up.

The lesson isn't that OpenAI's doomed. They're still enormously valuable, still shipping impressive products, still sitting on mountains of cash and talent. The lesson is that in AI, there are no permanent winners.

Google thought they'd dominate AI forever. Then ChatGPT showed up. OpenAI thought they'd revolutionised the industry. Then their competitors caught up.

Whoever's on top in December 2028? Your guess is as good as mine.

Key Takeaways

The Market Shift:

• Anthropic leads enterprise AI with 32% market share (OpenAI: 25%, Google: 20%)
• OpenAI's enterprise share has been cut in half since 2023
• ChatGPT user growth slowed to 6% while Gemini grew 30%

What's Driving the Change:

• Anthropic's reliability and safety-first approach resonates with enterprise buyers
• Google's Gemini 3 and Deep Think mode are getting strong reviews
• Claude Code hit $1 billion run-rate revenue in six months

For Business Leaders:

• Don't assume today's market leader will be tomorrow's
• Multi-model strategies reduce vendor lock-in risk
• Enterprise reliability often matters more than cutting-edge features
• Switching costs between AI providers are lower than expected

The Bigger Picture:

• AI competition is accelerating, not stabilising
• Three-year disruption cycles may become the norm
• The company that disrupts is often disrupted next

---

What to Watch

GPT-5.2 expected any day now

Reports suggested OpenAI accelerated GPT-5.2's release to early December, though no official announcement has materialised yet. If it arrives this week, it would be OpenAI's fastest model iteration yet, just weeks after GPT-5.1's November launch. (9to5Mac, iClarified)

Note: OpenAI hasn't officially confirmed any release date. The original reports cited industry insiders, not official sources. Delays would be unsurprising given the rushed timeline.

Internal priority escalation

OpenAI uses a colour-coded priority system internally. ChatGPT improvements were previously at "code orange" before Altman escalated to "code red," their highest level. This suggests the competitive threat is being treated as existential, not merely significant. (The Information)

Altman's counter-claim

In his internal memo, Altman reportedly claimed OpenAI's next reasoning model already beats Gemini 3 in internal benchmarks. If GPT-5.2 delivers on this, the "code red" narrative could flip quickly from crisis to comeback story. (WebProNews)

Note: Internal benchmarks are notoriously unreliable predictors of real-world performance. Independent testing will tell the real story.

The user number watch

ChatGPT's daily active users reportedly dropped from approximately 106 million to 100 million in late November. Whether GPT-5.2 reverses this trend, or whether Gemini continues its 30% growth trajectory, will determine who's winning by year's end. (Tom's Hardware)

---