This is Part 2 of 4 in the “The Soul of the Machine: Altman vs. Amodei and the Battle for AI’s Future” series.
The Super Bowl ad buy cost Anthropic more than some of its early funding rounds. When the creative team presented the final cut in a conference room at Anthropic’s San Francisco office, Daniela Amodei watched it twice and approved without hesitation. Dario watched it once and said nothing for a long time.
The spot was thirty seconds. It opened with a split screen: on one side, a cluttered interface full of pop-ups and banner ads, a chatbot pausing mid-sentence to recommend a mattress brand. On the other, a clean white screen, a simple conversation, a tool that did what you asked without trying to sell you something. The tagline landed with surgical precision: “Claude. The one without all the drama.”
It was funny. It was effective. It was ruthless.
“The one without all the drama” was not just a dig at advertising inside ChatGPT. It was a reference to OpenAI’s boardroom coup of 2023, to the chaos of Altman’s firing and reinstatement, to the organizational turbulence that had made OpenAI as famous for its internal politics as for its technology. It compressed years of corporate dysfunction into five words and a knowing smirk.
Dario turned to Daniela. “This is a departure.”
“From what?”
“From competing on substance.”
“It is substance, Dario. It’s a factual statement about our product. We don’t have ads. They do. We haven’t had a public meltdown. They have.” Daniela paused. “And it’s also a Super Bowl ad. It’s supposed to be entertaining. That’s how the format works.”
He knew she was right. He also knew what the ad represented: the moment Anthropic stopped being the company that was above the fray and became the company that spent millions of dollars proving it was above the fray. There was a contradiction in there, and he could feel it like a stone in his shoe.
He approved the ad anyway.
On February 2, a week before the Super Bowl, OpenAI launched Codex.
Altman had been anticipating this moment for months. Codex was designed to compete directly with Claude Code – the developer-focused product that had become one of Anthropic’s most important revenue streams, generating an estimated billion dollars in annualized revenue. Claude Code had built loyalty among software engineers with its technical precision and an interface that made coding feel collaborative rather than mechanical. It was, in Altman’s analysis, the single greatest competitive threat Anthropic posed.
Codex launched with aggressive pricing, undercutting Claude Code’s professional tier by twenty percent. The marketing emphasized speed, integration with existing development environments, and the backing of the world’s largest AI company. Brockman personally oversaw the rollout, appearing in developer forums and on livestreams with the energy of someone who had taken the competition personally.
“We’re not responding to anyone,” Brockman told reporters who asked if Codex was aimed at Anthropic. “We’re serving developers. Period.”
Nobody believed him. The timing, the pricing, the feature comparisons embedded in the launch materials – everything pointed at Claude Code. It was an act of targeted aggression dressed in the language of market expansion.
Altman watched the initial numbers with satisfaction. First-day signups exceeded internal projections by forty percent. Developer Twitter was electric. The reviews were strong.
Then Sunday arrived.
Super Bowl LX drew 127 million viewers. Anthropic’s first ad aired during the second quarter, and by the time the third-quarter spot ran – “The one without all the drama” – social media had already turned it into the most-discussed commercial of the night.
Altman was watching from his apartment in San Francisco. He had known about the ad buy – Anthropic’s purchase of Super Bowl airtime had been reported weeks earlier – but he hadn’t seen the creative. When the second spot aired, the one with the split-screen comparison and the five-word tagline, he set down his drink and stared at the screen.
His phone was already vibrating. Brockman’s message arrived first: This is war.
Altman’s initial reaction was controlled. He opened X and typed a response that was, by his standards, measured: Anthropic’s ads were “clearly dishonest,” he wrote, and the company was “basically an expensive product to rich people and enterprises.” He added that “wrapping yourself in the flag of safety while spending millions on attack ads is a choice.”
The post went up at 9:47 PM Eastern. By midnight it had been seen twenty million times.
What Altman had not anticipated was the velocity of the backlash – not against Anthropic, but against him. The tone of his response read as rattled to many observers. Journalists who covered the industry noted that Altman had never before responded to a competitor with this kind of heat. The word “testy” appeared in multiple headlines. TechCrunch ran a piece titled “Sam Altman Got Exceptionally Testy Over Claude Super Bowl Ads.”
For a man who had survived a boardroom coup with preternatural cool, the display of irritation was itself a story.
In his apartment across the city, Amodei watched the ads land and felt something he hadn’t expected: discomfort.
Not about the ads themselves – they were performing exactly as designed. The social metrics were extraordinary. Claude downloads spiked. The earned media coverage was worth many times the ad spend. From a competitive standpoint, the Super Bowl buy was the most effective single marketing action in Anthropic’s history.
The discomfort was about what the success revealed. You cannot simultaneously position yourself as the company that is above the fray and spend eight figures on Super Bowl ads designed to humiliate your rival. You cannot warn about the dangers of a death race while accelerating the competitive intensity. The ads worked because they were mean, and they were mean because meanness gets attention, and getting attention was now the point.
Daniela called him at eleven. “The numbers are incredible. App downloads are up three hundred percent from baseline. Earned media estimates are already at four hundred million impressions.”
“Good.”
“You don’t sound good.”
“I’m fine. The ads did what they were supposed to do.”
“But?”
Dario paused. “But the next time I stand at a podium and talk about competing on substance, someone is going to play the Super Bowl clip. And they’ll be right to.”
Daniela’s response was quick. “Then stop standing at podiums and start winning. The substance is winning. The safety mission requires market share. Market share requires people knowing we exist. People know we exist now.”
She was right. She was almost always right about these things. Dario thanked her and hung up and sat in the dark, thinking about the space between the company he had founded and the company it was becoming.
The following days were a controlled detonation across social media.
Brockman was the first to escalate. Where Altman’s response had been raw, Brockman’s was pointed. He posted a thread on X dissecting Anthropic’s revenue model, noting that Claude’s premium pricing made it “an AI for people who can afford it” while ChatGPT’s ad-supported free tier served hundreds of millions. “Dario lectures about access while building a luxury product,” Brockman wrote. “We actually serve the world.”
It was a sharp line, and it landed. The framing – populist OpenAI versus elitist Anthropic – became the dominant media narrative for the better part of a week. Fortune ran a cover story: “OpenAI vs. Anthropic Super Bowl Ad Clash Signals AI’s Trash Talk Era.” CNN, Reuters, Vanity Fair – every outlet that covered technology or culture or both found the story irresistible. Two companies worth a combined $880 billion, founded by people who had once worked together, now sniping at each other during the biggest television event of the year.
Business Insider published a piece titled “Anthropic Skewered OpenAI and Won the AI Super Bowl.” The Observer ran “Sam Altman and Dario Amodei Clash Over Ads and the Future of A.I.” The coverage treated the rivalry as entertainment, which bothered both men in different ways – Altman because he preferred to control the narrative, Amodei because he preferred there to be no narrative at all.
Behind the spectacle, the money was moving.
On February 13, Anthropic announced it had closed a $30 billion funding round at a $380 billion valuation. The round was led by a consortium that included sovereign wealth funds, major tech companies, and some of the same venture capital firms that held positions in OpenAI. The financial press noted the irony: investors were funding both sides of the war, hedging their bets on which vision of AI would prevail.
Amodei announced the round in a blog post that was, in tone and substance, the opposite of the Super Bowl ads. It was measured, technical, focused on Anthropic’s research roadmap and safety commitments. It did not mention OpenAI. It read like a document written by the CEO Amodei wanted to be – the physicist who built a company to solve hard problems – rather than the CEO the market was turning him into.
Daniela reviewed the post and added a single line: “This funding will accelerate our ability to build AI systems that are both capable and safe – because the world deserves both, not a choice between them.”
It was a competitive line disguised as a values statement. Dario recognized what she had done and left it in.
Altman processed the $380 billion valuation with the specific frustration of a man who had watched a rival close the gap. OpenAI’s own valuation hovered near $500 billion, but the distance was shrinking. Eighteen months earlier, Anthropic had been valued at a fraction of OpenAI. Now the ratio was closer than anyone had predicted.
He called a senior leadership meeting the day after Anthropic’s announcement. The agenda was ostensibly about Q1 priorities. The actual subject was the competitive situation.
“Their valuation is a signal,” Altman told the room. “It tells the market that there are two serious players in this space, not one. That changes how customers think, how regulators think, how talent thinks.”
“So what do we do?” someone asked.
“We do what we always do. We ship. Codex is already outpacing projections. The ad model launches in March. We have products in pipeline that they can’t match.” He paused. “But I also want to be honest about something. The Super Bowl thing – the way he framed us – that’s going to be a recurring tactic. He’s going to run the ‘we’re the safe ones, they’re the reckless ones’ play every chance he gets. And some people will buy it.”
Brockman leaned forward. “Then we make the counterargument. Every. Single. Time.”
Altman nodded. But later that evening, alone in his office, he pulled up Amodei’s essay – “The Adolescence of Technology” – and read it for the second time. The first time, he’d skimmed it, looking for competitive angles, places where Amodei had overreached or made claims that could be challenged. The second reading was slower, less strategic.
The essay was good. Better than good, if he was being honest. The section on nuclear technology – the argument that the period between capability and governance is when the most irreversible mistakes happen – was rigorous and difficult to dismiss. The analogy to AI was imperfect, as all analogies are, but it contained a truth that sat uncomfortably alongside Altman’s own conviction that speed was its own kind of safety.
He closed the laptop and looked out at the San Francisco skyline. Somewhere across the city, Amodei was probably working, or reading, or staring at his own version of this view and thinking about the same questions from the other side.
The philosophical war had become a commercial one. The commercial one had become a public spectacle. And now, as both companies quietly began assembling political operations – as lobbyists were retained, as super PAC conversations turned from hypothetical to concrete – the spectacle was about to become something neither man fully controlled.
On February 12, Anthropic announced a $20 million donation to Public First Action, a super PAC dedicated to electing candidates who supported robust AI safety regulation. The same week, reports surfaced that OpenAI had been quietly funding its own political organization, aimed at promoting frameworks that favored innovation and broad access.
The battle was about to move to Congress. And in Congress, the rules were different from anything either man had learned in a lab or a boardroom. In Congress, your arguments belonged to whoever was willing to use them, and the distinction between principle and positioning collapsed under the weight of other people’s agendas.
Neither Altman nor Amodei was ready for what that would cost.
This is a work of fiction. While inspired by real events and public information, all characters, dialogue, and internal deliberations are entirely imagined. Any resemblance to actual private conversations or decisions is coincidental.
