Andreessen Horowitz (a16z) is making aggressive bets on AI infrastructure beyond traditional hardware, backing both a financial marketplace for computing power and a specialized platform for scientific research. The venture firm led a $33 million seed round for Ornn, a startup creating trading tools for GPU (graphics processing unit) resources, while also participating in a $200 million seed round for Mirendil, an AI platform designed specifically for scientists and researchers.

Why Is a16z Betting on GPU Marketplaces?

Today, companies that need computing power to train and run AI models typically negotiate private deals for GPUs, which leads to unpredictable pricing and limited access. Ornn aims to change that by turning GPU power into a tradable commodity similar to oil or gold futures. The platform would let buyers and sellers trade compute futures contracts, helping organizations hedge against sudden cost spikes and gain better visibility into future expenses.

The $33 million funding round reflects growing investor confidence that compute will become a critical asset class. Early funding of $5.7 million from other investors already supported the idea before a16z's larger commitment, suggesting demand for such financial tools is accelerating. Smaller AI teams stand to benefit most, as they gain predictability in their largest operational cost. Over time, a stable compute market could encourage more efficient use of data centers and renewable energy sources, potentially lowering barriers for startups entering the AI space worldwide.

What Is Mirendil, and Why Does It Matter?

While Ornn focuses on the financial infrastructure of AI, Mirendil takes a different approach by targeting the application layer. The startup, founded by former researchers and engineers from Anthropic and Google, emerged from stealth mode with a $200 million seed valuation of approximately $1 billion. Rather than building consumer-facing chatbots, Mirendil is developing a platform that enables scientists, researchers, and research institutions to create specialized AI systems tailored to specific scientific disciplines.

The platform is designed for experts in fields such as biology, chemistry, physics, healthcare, and materials science to build and deploy AI models for complex scientific workflows. This reflects a broader industry shift away from general-purpose AI tools toward highly specialized systems capable of accelerating research and innovation across industries. Industry analysts believe AI-powered scientific discovery could become one of the most transformative applications of artificial intelligence over the next decade, helping researchers uncover new medicines, develop advanced materials, and solve complex global challenges faster than ever before.

How a16z Is Positioning Itself in the AI Infrastructure Race

• Compute Financialization: By backing Ornn, a16z is betting that GPU trading will become as essential to AI operations as oil futures are to energy companies, creating a new market for financial derivatives and risk management.
• Vertical AI Specialization: Mirendil's focus on domain-specific AI reflects a16z's recognition that general-purpose models alone cannot serve the needs of specialized research communities, creating opportunities for vertical platforms.
• Infrastructure Ecosystem Building: The two investments together suggest a16z is constructing a comprehensive infrastructure ecosystem, from the financial tools that manage compute costs to the application platforms that deploy AI for specialized use cases.

The scale of Mirendil's funding highlights the intense competition among investors seeking exposure to the rapidly expanding AI sector. Kleiner Perkins and chipmaking giant Nvidia also participated in the round, alongside a16z. Nvidia's involvement is particularly significant given the company's central role in powering modern AI development through its GPUs, which have become essential for training and deploying sophisticated AI models. The participation of a GPU manufacturer in a platform designed to optimize GPU allocation suggests confidence that compute optimization will drive significant value.

The $200 million raise provides Mirendil with substantial resources to attract top talent, expand research and development efforts, and accelerate the rollout of its platform to scientific organizations worldwide. As artificial intelligence adoption continues to accelerate globally, organizations are increasingly seeking AI solutions tailored to their unique requirements rather than relying exclusively on general-purpose models. This demand has created opportunities for startups focused on vertical AI applications, particularly in sectors where domain expertise is critical.

Mirendil's launch comes at a time when investment in artificial intelligence remains one of the strongest themes in global venture capital markets. Despite broader economic uncertainties, investors continue to deploy significant capital into companies developing foundational AI technologies and infrastructure. With an experienced founding team, substantial financial backing, and a mission centered on advancing scientific research, Mirendil enters the market positioned as one of the most closely watched AI startups of 2026.

Together, these two a16z-backed investments reveal a strategic shift in how the firm views AI's future. Rather than betting solely on large language models or consumer applications, a16z is building positions across the infrastructure stack, from the financial tools that manage compute allocation to the specialized platforms that deploy AI for high-value research. This approach suggests that the next wave of AI value creation may come not from the models themselves, but from the systems that make those models more accessible, affordable, and specialized for specific industries and use cases.