On May 6, 2026, Anthropic CEO Dario Amodei walked out onto a stage at his company’s developer conference in San Francisco and said something you almost never hear from a tech CEO: Growth is the problem.

Anthropic had planned for a 10-fold expansion. What it got was 80-fold growth in Q1, on an annualized basis. Revenue has crossed $30 billion, up from $9 billion at the end of 2025. The company is weighing a funding round at a reported $900 billion valuation – which, if it closes at those terms, would likely surpass OpenAI’s most recent post-money valuation of $852 billion. And yet, as Amodei told the audience that day, “I hope that 80-times growth doesn’t continue because that’s just crazy and it’s too hard to handle.”

He wasn’t being falsely modest. Demand for Claude has already created what Anthropic described as “inevitable strain on our infrastructure,” hitting reliability and performance during peak hours. Hours before Amodei took the stage, the company announced a deal with SpaceX – which, earlier this year, merged with xAI, the company behind the Grok AI models, now rebranded SpaceXAI – to take over the entire compute capacity at the Colossus 1 data center in Memphis, giving it access to more than 300 megawatts of capacity and 220,000 Nvidia GPUs.

The detail worth noting: xAI and Anthropic are direct competitors at the model layer. The fact that Grok’s infrastructure is now running Claude’s workloads is the clearest signal yet of how constrained high-end compute capacity has become. That’s a bridge built under emergency conditions, not a planned expansion.

So, why should SEO professionals, content marketers, and entrepreneurs care about Anthropic’s infrastructure problems? Because this story is actually about something much bigger than one company scrambling for server capacity.

This Has Happened Before

In 2011, I read I’m Feeling Lucky: The Confessions of Google Employee Number 59 by Douglas Edwards, who was Google’s first director of marketing and brand management. That’s when I learned how close Google came to buckling under its own success in the early days.

In late 1999, Edwards wrote, “Google began accelerating its climb to market domination. The media started whispering about the first search engine that actually worked, and users began telling their friends to give Google a try. More users meant more queries, and that meant more machines.” Then the machines became impossible to get. A global shortage of RAM hit at the worst possible moment, and Google’s system, as Edwards put it, “started wheezing asthmatically.”

That infrastructure crisis drove decisions that shaped the web for the next two decades. Google started filtering duplicate content – even non-malicious versions like printer-friendly pages – because every redundant page required adding hardware without improving user experience. The constraint shaped the product. The product shaped SEO.

Anthropic’s compute crisis is the same dynamic, playing out 25 years later at a different scale. The question isn’t whether they’ll solve it. They will. The question is what decisions they’ll make under pressure, and how those decisions will reshape the products that millions of marketers depend on.

What The Data Actually Shows

When I went looking for what this growth moment means for practitioners, I found the headlines and the data pointing in surprisingly different directions.

Rand Fishkin recently shared findings from the Datos State of Search Q1 2026 report, which draws on clickstream data from tens of millions of real devices. His summary was pointed: AI is disrupting traditional search – no, the data doesn’t show that. AI tools are growing faster than traditional search in absolute terms – no, traditional search is still outpacing AI tool growth on an absolute basis. AI Mode in Google is huge – no, it’s still under 0.2% share, growing but still small. ChatGPT is pulling away from Claude – actually, no. Claude is closing the gap, Gemini holds the number two spot and is growing, and ChatGPT has plateaued since September 2025.

These are not the narratives that get clicks. They are, however, what the data says.

At the same time, I went to Think with Google and worked through its report, “The Rise of the Super-Empowered Consumer,” which tells a different part of the same story. Some of what’s in there deserves more attention than it’s getting. AI Overviews is used by over 2 billion people, and users report making decisions faster and with more confidence. AI Mode now has over 75 million daily active users, with nearly 1 in 6 queries using voice or images. Queries in AI Mode run three times longer than traditional searches, and sessions are becoming more conversational. Google Lens handles over 25 billion visual searches every month. Shoppers are 2.3 times more likely to use Google Search than ChatGPT for purchase decisions, and 40% of consumers who use Google AI Mode for shopping say they’re using ChatGPT less as a result.

Two different pictures of the same moment. Both accurate. Neither is complete on its own.

The Takeaway For Practitioners

The AI industry is generating a firehose of information, and most of it gets consumed at the headline level. A company announces 80-fold growth, and people read it as a story about AI winning. Fishkin publishes data showing traditional search still outpacing AI tools in absolute volume, and people read it as a story about AI losing. Google publishes a consumer report showing AI Overviews reaching 2 billion users, and people read it as confirmation that SEO is dead.

None of those readings are wrong. All of them are incomplete.

The strategic value isn’t in reading the news. It’s in following the thread further – downloading the Datos report, working through the Google consumer study, checking the CNBC article against the Cryptopolitan analysis of what the Anthropic-SpaceX deal actually signals about the infrastructure war playing out between the major AI companies.

Google’s early infrastructure crisis produced lasting decisions about duplicate content that practitioners are still navigating. Anthropic’s current one will produce decisions about rate limits, model availability, enterprise pricing, and compute allocation that will shape how Claude-powered tools perform for the marketers and developers using them. Those decisions are already being made.

The practitioners who understand the context those decisions come from will be better positioned than those who only read the headline.

More Resources:

Featured Image: Anton Vierietin/Shutterstock