Nvidia’s recent leadership change and strategic shift may open the door for emerging chipmakers, particularly in Europe, to challenge the company’s dominant position in the artificial intelligence (AI) processor market. Jensen Huang, Nvidia’s former CEO, established a powerful ecosystem by providing financial incentives to customers purchasing Nvidia chips, combined with the company’s proprietary Cuda programming platform. This approach helped Nvidia capture close to 90 percent market share in several segments.

However, Huang’s departure and the appointment of interim chief financial officer Colette Kress, who is adopting a more cautious approach to investments, could slow Nvidia’s aggressive ecosystem expansion and weaken the competitive edge provided by the Cuda platform. As a result, alternative chip providers may find it easier to attract clients and gain market share.

This shift is already reflected in the strategies of companies like Anthropic, which recently stated in its listing prospectus that it could onboard new chip platforms far more quickly than before, reducing its dependence on Nvidia by sourcing from other providers such as Amazon Web Services, Google, and AMD.

Alongside these established alternatives, a number of smaller startups—many based in Europe—are entering the market. For instance, UK-based Fractile recently raised funds at a $7 billion valuation, while Olix, founded by James Dacombe, the youngest self-made billionaire, is valued at $3.3 billion and counts SoftBank-backed Arm as a backer.

These startups are betting on two key industry trends. First, a move away from general-purpose processors toward specialized chips tailored for specific tasks, allowing companies to deploy different types of chips depending on their computational needs. Second, by designing chips from scratch, newcomers aim to better meet the rapidly escalating computing demands of modern AI models.

Technical innovation is central to this effort. Nvidia and its competitors currently use two types of memory chip designs: one where memory is separated from the processor, providing larger capacity but slower speeds, and another with memory embedded inside the processor, offering faster operation but limited capacity. Overcoming this trade-off is critical as AI models grow larger and more complex. For example, Fractile is developing technology that places dynamic random-access memory (DRAM) directly on top of the processor to enhance both speed and capacity.

Other startups, including Callosum—backed by the UK sovereign venture fund—are focusing on systems that manage and orchestrate diverse chip usage.

Despite the potential, several challenges remain. Some customers appreciate the concept of task-specific chips, while others are concerned about potential inefficiencies, such as idle time when specialized chips are not in use. Additionally, many of these advanced chips may not reach production before 2028, by which time the economics and technology landscape of AI could significantly evolve.

Commercializing new chip designs also carries risks. Past examples include Graphcore, a British company once valued at $2.8 billion that was eventually acquired by SoftBank for substantially less.

Ultimately, Nvidia’s leadership change could mark a turning point in the AI chip industry, demonstrating that the dominance of its Cuda platform is not unassailable. This shift creates opportunities for a broader range of chipmakers to compete and innovate in a rapidly evolving market.