Nvidia invests $150 million in Baseten as AI industry shifts from training to inference

Nvidia Backs AI Inference Startup in Strategic Shift from Model Training to Inference

Nvidia has committed $150 million to Baseten, an AI inference startup that just closed a $300 million funding round at a $5 billion valuation, more than doubling its previous worth1

1

. The round was co-led by Institutional Venture Partners and CapitalG, Alphabet's independent growth fund, marking a significant milestone for the San Francisco-based company founded in 20192

2

. Nvidia's investment underscores the chip giant's increasingly aggressive push into startups focused on AI inference—the process of running trained machine learning models in production environments to generate real-world outputs3

3

.

The Shift from Model Training to Inference Accelerates

This deal highlights a fundamental transformation across the AI ecosystem as attention pivots from building ever-larger models to deploying them efficiently at scale. Nvidia CEO Jensen Huang has repeatedly emphasized that AI inference will ultimately become a much larger market than model training1

1

. As enterprises move from experimentation to large-scale deployment, demand for reliable and cost-efficient infrastructure is accelerating, placing companies like Baseten at the center of this transition. The investment also represents another instance of Nvidia backing a direct customer of its AI chips, a pattern that has raised some industry concerns2

2

.

Baseten's AI Infrastructure Platform Powers Major Applications

Baseten helps companies deploy large language models and run them in production with minimal friction. The AI infrastructure platform serves notable clients including AI code editor Cursor and note-taking platform Notion, powering applications that reach hundreds of millions of users4

4

. Co-founder and CEO Tuhin Srivastava has described Baseten's ambition as building the "AWS for inference," providing the infrastructure that allows AI companies to take performance and reliability for granted1

1

. The platform offers tooling, orchestration software, and optimized runtime environments that maintain consistently low latency and high availability even under heavy load2

2

.

Optimized for Nvidia's Latest GPU Architectures

Baseten's platform is specifically optimized for Nvidia's latest GPU architectures, including the H100 and next-generation B200 chips1

1

. By enabling high-performance inference workloads on these GPUs, Baseten effectively extends Nvidia's ecosystem, helping ensure its hardware remains the default choice as AI adoption spreads across enterprises. The company also offers Truss, an open-source framework that simplifies model deployment by allowing teams to package models, manage dependencies, and scale inference workloads with minimal configuration—an increasingly critical capability for AI application developers1

1

.

Rapid Growth Signals Market Momentum

The latest funding marks Baseten's third raise in just 12 months, following a $75 million Series C in May and a $150 million Series D in September4

4

. With the new capital, the company has now secured $585 million in total funding1

1

. Srivastava told PYMNTS that every organization will either become AI-first or AI-enabled, and companies can only move fast if they delegate away model orchestration that isn't core to their business4

4

.

Nvidia's Broader Inference Strategy Takes Shape

Nvidia's investment in Baseten is part of a larger strategic push into inference technology. The company holds $60.6 billion in cash, cash equivalents, and marketable securities as of October 26, 2025, providing substantial resources for this strategy3

3

. Recent moves include licensing inference technology from Groq in December, committing to invest up to $100 billion in OpenAI, and taking stakes in dozens of smaller companies developing technology for AI applications4

4

. Nvidia has also pursued acquisitions to secure top AI talent, with advanced talks to acquire AI21 Labs in a deal valued between $2 billion and $3 billion, primarily to access the startup's roughly 200 specialized machine-learning engineers. This follows a recent $20 billion agreement with Groq centered on transferring elite talent and gaining access to language processing unit technology. As AI features become embedded directly into consumer and enterprise products across sectors like productivity software, finance, and creative tools, investors argue that inference platforms are well-positioned to capture long-term value1

1

.