Nvidia's web of AI bets: smart strategy or bubble risk?

Nvidia has recently poured $3 billion into Meta's share sale, joining a web of AI-related investments that has made the chip giant a key player in the artificial intelligence boom. These investments include stakes in leading AI developers like OpenAI, Anthropic, and Elon Musk's xAI, as well as neocloud companies such as CoreWeave and Nebius.

Nvidia's involvement extends globally, with backing for French AI developer Mistral, British self-driving startup Wayve, and a partnership with South Korea's SK Group to build data centers.

The company's financial backing extends beyond equity, with around $36 billion in cloud backstops and a $105 billion guarantee linked to an OpenAI data center. Experts argue that Nvidia's extensive profits lead it to channel funds back into AI infrastructure, making it a smart move. However, critics question whether Nvidia's investments are contributing to an AI bubble.

While Nvidia defends the approach, arguing that frontier AI labs need significant funding to succeed, concerns persist about the concentration of risk in these relationships.

Two direct customers, CoreWeave and Nebius, account for 22 percent and 14 percent of Nvidia's revenue, respectively. If AI demand disappoints, this concentration could lead to a domino effect, impacting chipmakers, data-center builders, and their financiers. Some experts expect high demand for Nvidia chips to continue for three to five years, despite supply limitations.