"우리는 코드 레드를 하지 않습니다": Anthropic CEO, OpenAI 및 Google 비판

The AI industry is often portrayed as a breathless sprint, a race where speed is the only metric that matters. But at the New York Times DealBook Summit this Wednesday, Anthropic CEO Dario Amodei offered a different perspective—one that positioned his company not as the fastest runner, but as the only adult in a room full of panicked gamblers.

In a wide-ranging interview with Andrew Ross Sorkin, Amodei delivered a stinging critique of his biggest rivals, mocking OpenAI and Google for their tendency to declare “code red” emergencies and warning that the industry’s current financial trajectory is bordering on reckless.

The “Code Red” Culture

For the last three years, the narrative of Generative AI has been defined by reaction. When ChatGPT launched in late 2022, Google famously declared a corporate “Code Red,” reshuffling its entire internal structure to prioritize AI safety be damned. More recently, reports surfaced that OpenAI entered a similar crisis mode following the launch of Google’s Gemini 3 capabilities.

Amodei finds this hilarious.

“We don’t have to do any code reds,” Amodei stated flatly, highlighting a distinct difference in company culture. “There are some players who are YOLO, who pull the wrist dial too far.”

By framing the competition as reactive, Amodei is making a subtle but devastating point: Panic is not a strategy. When companies exist in a perpetual state of emergency, they make mistakes. They release models that aren’t ready, they burn cash inefficiently, and they confuse motion with progress. Anthropic, by contrast, claims to move with “high velocity” but without the hair-on-fire desperation.

The “YOLO” Economy: A Warning on Financial Suicide

Perhaps more cutting than his cultural critique was his financial one. Amodei warned that the current level of capital expenditure (CapEx) in the AI industry—specifically the hundreds of billions being poured into data centers—is disconnected from economic reality.

“There is a risk of overextension,” Amodei warned. “Some are turning the dial too far.”

His argument is simple math:

1. The Spend: Tech giants are building infrastructure based on the assumption that AI revenue will explode to match it immediately.
2. The Assumption: To justify a $50 billion data center, you need to believe that a single AI model will generate tens of billions in revenue per year very soon.
3. The Reality: While Anthropic itself is projecting robust growth—potentially hitting $10 billion in revenue in 2024—Amodei cautioned against assuming that number will magically jump to $200 billion overnight.

“If you build the infrastructure for a $200 billion business, but only have a $20 billion business, you have a solvency problem,” he implies. This “YOLO” approach to balance sheets is what sets the current bubble apart from steady growth. Amodei suggests that while others are betting the entire company on the next model being ‘God-like’, Anthropic is building a sustainable business based on the technology that exists today.

Enterprise Focus vs. Consumer Hype

Why can Anthropic afford to be so calm? Because they aren’t fighting the same war.

While OpenAI and Google are locked in a death match for consumer eyeballs—fighting over which chatbot has the best voice mode or can generate the funniest images—Anthropic has quietly cornered the market that actually pays: The Enterprise.

“We’ve optimized our models more and more for the needs of businesses,” Amodei explained.

This strategy protects them from the fickleness of consumer trends. A teenager canceling a $20/month ChatGPT subscription because they got bored is a high-churn problem. A Fortune 500 bank integrating Claude into their entire coding workflow is a sticky, long-term revenue stream.

By focusing on high-value, complex applications like scientific reasoning and advanced coding, Anthropic builds deep moats. “Code Reds” happen when you are terrified of losing users to a shiny new toy. When your product is deeply embedded in corporate infrastructure, you don’t need to panic every time a competitor releases a demo video.

The Verdict

Dario Amodei’s comments at DealBook 2024 mark a maturing of the AI narrative. For years, the story was “Move Fast and Break Things.” Amodei is suggesting that the new winning strategy might be “Move Deliberately and Build Things That Work.”

If OpenAI and Google are the rockstars trashed hotel rooms and living fast, Anthropic is positioning itself as the investment banker who actually owns the hotel. And as the bill for the AI boom starts to come due, the one who “doesn’t do code reds” might be the only one left standing.