Anthropic executive takes a thinly-veiled swipe at OpenAI over spending and ads

Anthropic is focused on growing its business rather than making “flashy headlines,” its commercial chief told CNBC in a thinly-veiled swipe at rival OpenAI as the public war of words between the AI giants continues.

Anthropic aired ads at Sunday’s Super Bowl taking a dig at OpenAI’s decision to begin testing ads on ChatGPT. OpenAI CEO Sam Altman called Anthropic’s ad “deceptive.” It comes as the leading AI model players are locked in an intensifying battle to sign up businesses to use their products.

In a wide-ranging interview with CNBC, Smith also said the market sell-off in software stocks, sparked by Anthropic’s Claude Cowork tool, was “a lot of hyperbole.”

Anthropic spent millions on its Super Bowl commercials, a 60-second pregame ad and a 30-second in-game ad, which stressed said: “Ads are coming to AI. But not to Claude.”

Smith told CNBC it was a “conscious decision” not to include ads in Claude. Advertising would take Anthropic in “directions where you’re optimizing for the wrong things,” he said. He added that, without ads, the company can focus on areas such as making AI models more intelligent and being “genuinely helpful, safe, and trusted.”

While OpenAI has a large consumer focus with ChatGPT, Anthropic has focused on selling its AI to businesses.

Smith said Anthropic is “unconflicted” by not offering ads, as it focuses on selling its AI to businesses.

“We’re not fighting another partner for eyeballs or for ad revenue or anything,” he added.

“Our focus is on model quality, model efficacy, how it can integrate across the wider enterprise, which comes back to all of the investments that we’re really focused on… we are exclusively focused on a different set of things. Our attention is not split.”

Altman said Anthropic’s Super Bowl ads were “funny” but “clearly dishonest,” adding that OpenAI would “obviously never run ads in the way Anthropic depicts them.”

OpenAI had not responded to a CNBC request for comment as this article went live.

Spending on infrastructure has been a key focus for investors after the latest Big Tech earnings season, which saw companies from Alphabet to Amazon ramp up their capital expenditure plans for 2026.

Anthropic has committed $50 billion to building data centers in the U.S., but it also spends money buying compute from players like Microsoft and Google. Meanwhile, OpenAI has committed more than $1 trillion to future infrastructure developments with partners including Nvidia, Oracle and Broadcom.

When asked about Anthropic’s approach to infrastructure compared with companies like OpenAI, Smith said: “We’ve made less flashy headlines than some, and we’ve been focused on growing revenue and winning business, rather than spending money and announcing the biggest compute deals that we possibly could.”

OpenAI announced a string of partnerships in recent months. Among them, Nvidia said it would commit $100 billion to support OpenAI as it builds and deploys at least 10 gigawatts of Nvidia systems.

More recently, OpenAI announced a $10 billion deal with chipmaker Cerebras to deploy 750 megawatts of Cerebras’ AI chips. It has other agreements with AMD and Broadcom.

Meanwhile, Anthropic’s CEO Dario Amodei has reinforced a “do more with less” mindset and said the company’ i taking a more disciplined approach to spending than others.

Smith said Anthropic’s leadership is discussing how much to spend on compute daily. But he added that he was “comfortable” with its spending because demand remains strong.

“This isn’t us buying ahead of demand,” Smith said, praising the “incredible growth” in Claude Code and Cowork, two key products Anthropic sells to businesses. Smith added there had been “incredible growth in enterprise business overall, that we absolutely need to satisfy.”

Smith added that the company is constantly reviewing its spending commitments. “We are looking at buying as close to the right amount of compute to keep us on that very, very significant acceleration curve, and not go too much, not go too little, because too little would be bad for our customers,” he said, adding that there would be more infrastructure announcements made soon.

Smith was speaking after Anthropic on Wednesday struck a partnership with Man Group, an investment management firm that will use Anthropic’s AI products. The companies will also build new tools together.

Last week, software stocks took a hammering after Anthropic’s productivity tool Claude Cowork gained traction. Investors feared that AI could do more of the processes that companies were paying multiple software vendors for.

When asked if enterprises are looking at using Anthropic’s products to phase out the software they were traditionally using, Smith said it depends on the organization.

Some organizations are “doubling down” on software and others are looking at tools like Anthropic’s, Smith told CNBC.

There was “a lot of hyperbole in the market last week,” Smith said.

“As a lot of folks pointed out, those applications performed some very important tasks in an enterprise,” Smith said, referring to current software that’s used in organisations.

“They have some very specific data models. They have some very specific workflows that organizations are going to get a lot of value from for a very long time.”