ZML Launches Free LLMD Software to Break Nvidia, AMD, and Google TPU Vendor Lock-in for AI Inference

On July 8, 2026, French AI startup ZML released ZML/LLMD, a free multi-chip AI inference software designed to break vendor lock-in for large language models by enabling them to run at maximum speed across diverse hardware, including Nvidia, AMD, Google TPUs, Apple Metal, and Intel Arc.

Addressing AI Inference Challenges

The ZML/LLMD software is engineered to support various AI chips, including those from Nvidia, AMD, Google TPUs, Apple Metal, and Intel Arc. By providing compatibility across these different architectures, ZML seeks to empower organizations to utilize their existing hardware more efficiently and avoid reliance on a single chip manufacturer. This approach could significantly reduce the operational costs associated with AI inference, a critical factor for large-scale deployments.

Endorsement and Market Impact

The initiative has garnered support from prominent figures in the AI community, including Turing Award winner and Meta chief AI scientist Yann LeCun. LeCun's endorsement highlights the potential of ZML's technology to foster a more open and competitive AI ecosystem. The inference market, often referred to as the "inference gold rush," represents a substantial segment of the AI industry, with companies like Baseten recently valued at $13 billion. ZML's entry into this space with a free, multi-chip solution could disrupt established dynamics and offer new opportunities for innovation.

Empowering European AI Chip Startups

ZML's chip-agnostic ZML's chip-agnostic platform could particularly benefit Eu could particularly benefit European AI chip startups. Companies such as Axelera, Fractile, Kalray, and SiPearl, which are developing their own specialized AI hardware, may find it easier to compete against dominant players like Nvidia if their chips can seamlessly integrate with widely adopted inference software. This could foster greater diversity in the AI hardware landscape and promote regional technological independence.

Practical Implications for Enterprises and Cloud Providers

For enterprises and cloud providers, ZML/LLMD offers a pathway to greater flexibility and cost efficiency in their AI operations. The ability to run LLMs optimally across heterogeneous infrastructure means that organizations are not confined to specific hardware vendors, allowing them to choose solutions based on performance, cost, or availability. This freedom from vendor lock-in is a key driver for many businesses investing in AI technologies.

Conclusion

ZML's release of ZML/LLMD on July 8, 2026, marks a notable step towards democratizing AI inference by offering a free, multi-chip compatible server for large language models. With its potential to reduce vendor lock-in and support diverse hardware, this software could reshape how enterprises and cloud providers manage their AI infrastructure, fostering a more competitive and cost-effective environment for AI development and deployment.

Sources

• https://techcrunch.com/2026/07/08/hot-french-startup-zml-releases-free-product-to-speed-inference-across-lots-of-ai-chips/
• Microsoft announces powerful new chip for AI inference - TechCrunch
• What is Mistral AI? Everything to know about the OpenAI competitor
• Station F ramps up as a launchpad for Europe's hottest AI startups
• Mistral is rumored to be raising €3B at €20B valuation | TechCrunch