Anthropic is talk of the town
Samuel Boivin | Nurfoto | Getty Images
If one thing became clear at this week’s HumanX conference in San Francisco, where 6,500 executives, founders and investors gathered to talk about artificial intelligence, it’s that OpenAI no longer dominates the industry conversation. At least for now, that distinction belongs to Anthropic.
Although many participants openAI, Cursor and Google offers powerful alternatives.
Despite its feud with the Pentagon, which became public last month and quickly found its way into the courtroom, Anthropic has only gained momentum. The Department of Defense blacklisted Claude, but after appealing the rulings of two courts, Anthropic can continue to work with other federal agencies while the cases are ongoing.
Anthropic’s initial strength in the organization positioned it to benefit from the growing popularity of AI coding agents used to create, edit and review code. As OpenAI kicks off the generative AI boom with the launch of ChatGPT in 2022, Anthropic may be best set to win the contracts of the biggest spenders.
CNBC interviewed 19 executives and investors at HumanX; Some of them requested anonymity so they could speak freely. Here are my top three takeaways.
Claude ‘became a religion’
Anthropic was founded in 2021 by a group of researchers and executives who left OpenAI. Startup is valued $380 billion, making it one of the most valuable private companies in the world.
Claude Code went public in May 2025 and was generating more than $2.5 billion in annual revenue as of February. Arvind Jain, CEO of enterprise AI company Glean, said Claude Code inspired “Claude Mania,” which put pressure on business leaders to implement it.
“This has become a religion, this is the level of madness,” Jain said in an interview. “Folks, today I can go and ask them, ‘Hey, if I gave you an AI tool, which tool would you want?’ The answer will be Claude.”
On Tuesday, Anthropic announced Claude Mythos Preview, a new AI model with advanced cybersecurity capabilities thanks to strong coding and reasoning skills. The model has generated a lot of interest at HumanX, although its launch was limited to a select group of approximately 50 companies.
Victor Riparbelli, CEO of AI video company Synthesia, said Anthropic has managed to show focus and restraint with its models and product, which can be difficult for a young hyper-growth company.
“The guys at Anthropic said, ‘We’re not going to do anything about the video, we’re not going to care about the audio models, we’re just going to decode gen,’ and now here we are,” Riparbelli said in an interview. “OpenAI faced the problem of having to market six different products, which took up space in the consumer’s mind.”
One investor cautioned that Anthropic has been consistent and managed to identify a compelling AI use case, but the industry is still young and momentum could easily shift in another direction.
AI change management
As tech companies try to usher their customers into the age of artificial intelligence, they are also grappling with how to use and deploy agents internally. Keeping up with the pace of change is no easy task, even for startups in Silicon Valley.
Ashwin Sreenivas, president of AI startup Decagon, said the emergence of coding agencies has led to a number of changes at his company. Decagon changed its interview process to allow candidates to use the tools, and the company can rely on smaller teams of engineers.
On a project that might require four or five engineers, “everyone becomes two engineers because they can move much faster and go much further,” Sreenivas said in an interview.
For Navrina Singh, CEO of AI governance startup Credo AI, the proliferation of new AI tools has been both exciting and worrying for her. He said over-communicating, especially with customers, has become important.
“Things that I couldn’t do last year and had to hire 10 people, I can actually build in a weekend and distribute them for myself and the company,” Singh said. “My concern is that I can’t control my roadmap and I can’t control my commitments to enterprise customers who like more clarity and a little more stability.”
Large technology companies are also moving towards similar changes.
Cisco President Jeetu Patel said about 85% of his company’s engineering workforce, or about 18,000 employees, are using AI, but the path to get there is different than he expected. Cisco learned that in the beginning, I needed to prioritize adoption over results and trust that the model capabilities would continue to improve, Patel said.
“You can’t think of these as tools; you have to think of them as digital colleagues joining your team, because the composition of your scrum team changes,” Patel said at the conference. “You might not have a scrum team of eight people. You might have a scrum team of two people and six reps, or two people and infinite reps.”
Race against China
Qwen3 is Alibaba’s latest big language model that is said to combine traditional LLM capabilities with “advanced, dynamic reasoning.”
Stick Pictures | Light Rocket | Getty Images
Instead, they focus on another looming geopolitical issue: China’s deficit-heavy models.
In AI, if a model’s parameters or elements that improve its outputs and predictions during training are publicly available, that model is considered open weight. Chinese open weight models, including GLM-5.1, Kimi K2.5 and Qwen3.5 as of April Mastering industry benchmarks.
American companies are flocking to Chinese models. Cursor built it Composer 2 model I use some 2.5. Airbnb CEO Brian Chesky told CNBC in October that his company’s chatbot is heavily dependent on Alibaba’s Qwen.
Given the emphasis the U.S. AI industry is placing on beating China in innovation, there is a great deal of emphasis on closing the gap in open gravity domestically. Two investors told CNBC they are dedicating most of their time and resources to the effort, and a third said it’s one of the key problems the industry needs to solve right now.
Glean’s Jain said having multiple options is critical.
“The trend we are seeing is that businesses today are very cautious about sticking to one or two providers for all their AI,” Jain said. “They don’t want to work with one model company because they know innovation happens across many companies and also in open source. You want to have a choice.”
WRISTWATCH: OpenAI criticizes Anthropic in note to shareholders
