Just a year ago, Inflection AI was among the most prominent startups, releasing top-tier AI models it claimed could outshine those from OpenAI, Meta, and Google. Today, however, the company has shifted its focus, with its new CEO revealing that Inflection is no longer aiming to compete in creating next-generation AI models.
This pivot follows significant changes within the company. Microsoft hired former CEO Mustafa Suleyman to lead its AI business, paying $650 million to acquire much of Inflection’s staff and license its technology. In recent months, Inflection limited usage of its consumer AI chatbot, Pi, and started focusing more on enterprise clients.
On Tuesday, Inflection announced that in the past two months, it had acquired three AI startups to enhance its enterprise-focused offerings. The company is also open to licensing AI models from its former competitors in the future.
Meanwhile, the Federal Trade Commission is reportedly investigating Microsoft’s partial acquisition of Inflection to determine if the deal reduces competition.
Sean White, who took over as CEO after the Microsoft deal, emphasized that while Inflection has stepped away from competing in frontier AI model development, it remains competitive in the enterprise space.
“I don’t see the need to compete with companies building the next 100,000-GPU systems,” White told TechCrunch, referencing the few highly funded firms working on cutting-edge AI models, including Microsoft.
He clarified: “When I say we can’t compete with them, it’s not just about capacity—it’s about choosing not to. However, we are still very much competing with them in the enterprise space. Our focus is on building tools and solutions tailored to meet enterprise needs, which sets us apart.”
White believes that today’s AI models are more than sufficient to meet the needs of most enterprises. He’s also skeptical about the practicality of test-time compute scaling, often hailed as the hallmark of next-generation AI models. According to White, AI labs have cleverly marketed high latency as “thinking” to make consumers more accepting of their models’ limitations.
“There’s a part of me that laughs at how latency in inference is now being spun as ‘thinking,’ rather than admitting these systems are just getting bigger and slower,” White remarked.
Rather than chasing cutting-edge AI research, Inflection is now focusing on providing practical AI tools for enterprises. On Tuesday, the company announced the acquisition of two startups as part of this shift: Jelled.AI, which uses AI to streamline employee inboxes, and BoostKPI, which specializes in AI-powered data analytics. Last month, Inflection also acquired Boundaryless, a European automation consulting firm, to strengthen its overseas operations.
While Inflection currently uses its own models, White hasn’t ruled out leveraging other AI models in the future. A key selling point for Inflection is its ability to run AI on-premises, a significant advantage for enterprises prioritizing data security compared to cloud-based solutions from major AI labs.
These acquisitions have allowed Inflection to bolster its talent pool and expand its product offerings. However, competition in the enterprise AI space is fierce. Industry leaders like Salesforce are heavily investing in AI agents, while Meta has launched a new business-focused AI division. Startups such as Anthropic and Cohere are also building enterprise-specific AI solutions. Despite this, Inflection believes it is better positioned to succeed in the enterprise market than to compete with frontier AI labs pushing the boundaries of AI capabilities.
Leave a Reply