NVIDIA shifts from selling chips to investing in ecosystems: what do three AI investments in one day reveal

AI inference is becoming NVIDIA’s new strategic focus. According to the latest news, NVIDIA has invested $150 million in AI inference startup Baseten. This round of financing for Baseten totals $300 million, with the valuation doubling to $5 billion. More notably, within the same time frame, NVIDIA also made significant investments in two AI startups, Humans and Anthropic, reflecting a profound shift in NVIDIA’s strategic approach.

Inference Becomes a New Industry Focus

Baseten specializes in AI model inference and deployment, an area that is often overlooked but extremely important. If training is the “learning phase” of an AI model, inference is the “application phase”—the process by which the model generates outputs based on inputs.

This round of funding was led by venture capital firms IVP and Alphabet’s independent growth fund CapitalG, with NVIDIA participating as a strategic investor. The investment portfolio is quite interesting: it includes both professional VCs and tech giants, indicating that the inference field has shifted from a niche topic to an industry consensus.

NVIDIA’s Ecosystem Investment Map

What better illustrates the point than NVIDIA’s investment pace? In just the past few days, NVIDIA has appeared in multiple AI startup funding rounds:

This is not scattered investment but a systematic ecosystem layout. Humans was founded by former researchers from Anthropic, xAI, and Google, focusing on developing AI that enhances human collaboration; Anthropic is a major competitor to OpenAI, competing in the large model arena; Baseten is deeply engaged in the specific application scenario of inference deployment.

From “Buying Shovels” to “Buying Outputs”

According to industry analysis, AI investment is undergoing a significant shift by 2026: moving from simply “buying shovels” (investing in computing power and chips) to “buying outputs” (investing in applications and decision-making processes).

NVIDIA’s series of investments exemplify this shift:

• Past logic: AI companies need computing power, so buy NVIDIA chips
• Current logic: NVIDIA not only sells chips but also invests in companies using those chips, ensuring ecosystem prosperity

The strategic reasoning behind this is clear. As the AI industry moves from infrastructure building to application deployment, pure chip suppliers risk being squeezed out. NVIDIA, by investing in key players across the ecosystem, can both promote chip demand and gain returns from applications, while also strengthening its influence over the industry chain.

Why Now

The industry’s shift in focus is the direct trigger for this transformation. NVIDIA CEO Jensen Huang has previously emphasized that as AI moves from model training to large-scale inference and deployment, the high inference costs and complex workflows have become new bottlenecks. Companies like Baseten, which optimize inference, and Humans, which integrate AI applications, directly address this pain point.

Meanwhile, NVIDIA is also strengthening control over the upstream industry chain. Reports indicate that Jensen Huang has invested heavily to secure TSMC’s capacity, demonstrating that NVIDIA is not only ensuring chip supply but also deepening its ties across the entire industry chain.

Summary

NVIDIA’s series of investments mark a turning point: from being a pure chip supplier to becoming an AI ecosystem builder. By investing in startups across inference, applications, and large models, NVIDIA is locking in chip demand while also building its moat. This is not just a strategic business adjustment but a judgment on the development stage of the AI industry—once infrastructure becomes stable, the next growth engine will be ecosystem prosperity and application deployment. For those seeking to understand the future of AI, observing NVIDIA’s investment moves is often more meaningful than watching stock prices.