Last week, over 3,000 attendees gathered in Las Vegas for the first-ever HumanX AI conference. The main theme of the event was trust — how to build reliable outcomes using powerful but unpredictable AI technology.

This was also the same week Apple decided to delay the launch of its AI features due to concerns about accuracy. At HumanX, the statistic that “only 6% of AI projects make it to production” flashed across the screen, emphasizing the challenges AI companies face. Despite this, investment in AI reached over $100 billion in 2024, an 80% increase from the previous year, according to the HumanX and Crunchbase AI report.

The event featured panels, product launches, Q&A sessions, and demos spread across multiple stages and rooms, as well as networking opportunities through a dedicated app. Among the wide range of topics covered, I found the perspectives from leading AI model companies especially fascinating, particularly as generative AI continues to evolve since the launch of OpenAI’s ChatGPT just over two years ago.

OpenAI’s Competitive Edge

OpenAI, valued at $157 billion, faced questions about its competitive advantage, especially from CNBC’s Kate Rooney, who asked, “What is your moat?” Kevin Weil, OpenAI’s Chief Product Officer, responded by acknowledging that the days of a one-year lead in AI development are over. However, he emphasized that even a three- to six-month lead is still incredibly valuable.

Weil also pointed out how rapidly AI technology advances: “Every two months, there’s a new model that can do something previously thought impossible for computers.” Despite the fast-paced competition, OpenAI remains dominant. Weil shared that over 3 million developers use the OpenAI API, 400 million people use ChatGPT each week, and more than 2 million businesses rely on OpenAI’s enterprise products.

Anthropic and the Future of AI Apps

A key discussion at the conference was between Alex Heath from The Verge and Mike Krieger, Chief Product Officer at Anthropic, about building an AI model company and their plans to create AI applications. Anthropic’s Claude Code, which launched just weeks ago, quickly gained 100,000 users in its first week.

Krieger explained how they reached out to Anthropic’s code API customers ahead of Claude Code’s release, putting it in direct competition with other major AI companies like GitHub’s Copilot and Codeium’s Windsurf. He stressed the importance of offering first-party products, as they provide crucial feedback that APIs alone cannot.

On a more philosophical note, Krieger shared his vision for the future of AI. He stated, “If it’s just chat boxes and chatbots a year from now, we’ll have failed.” His goal is to shape the future of human-AI interaction.

Mistral: Open Source and Smaller AI Models

Mistral AI, based in France, takes a different approach by focusing on open-source models to promote decentralized AI development. Mistral’s CEO, Arthur Mensch, highlighted the growing demand for open-source solutions, particularly for organizations with data governance and sovereignty needs.

Mistral not only offers open-source models but also a platform for deploying and managing AI applications. Their smaller models are particularly useful in robotics, with partnerships like the one with Helsing for drone technology. Mistral’s software stack is designed for specific hardware applications, which is crucial as AI becomes more embedded in robotics.

Though Mistral began as an enterprise-focused company, Mensch noted that once APIs are established, a product is not far behind. In fact, Mistral launched its first consumer product, Le Chat, just last month.

Looking Ahead to Next Year

Next year, the HumanX AI conference will be held in San Francisco, where most AI investment is concentrated. With nearly 30% of the companies presenting at the conference expected to be potential acquisition targets, it’s likely the landscape of AI companies will look quite different a year from now.