Even as Nvidia announced its $100-billion investment in OpenAI on September 22, industry stalwarts are warning against a dotcom bubble-like scenario in which companies fund their customers to drive demand.
On September 22, 2025, Nvidia announced its $100 billion investment in OpenAI to build 10 GW of AI data centres, a scale that could require 4 to 5 million GPUs—nearly double of Nvidia’s annual output. CEO Jensen Huang called the deal “monumental in size”, marking a pivotal moment in the industrialisation of artificial intelligence (AI).
In a press release announcing the deal, OpenAI CEO Sam Altman said: “Compute infrastructure will be the basis for the economy of the future, and we will utilise what we’re building with Nvidia to both create new AI breakthroughs and empower people and businesses with them at scale.”
During his conversation with Michael Milken, chairman of the economic thinktank Milken Institute's 2025 Global Conference, Huang said: “…for the very first time in history, we can imagine the opportunity to close that gap to put 30-40 million workers back into the workforce that otherwise the world doesn't have.” He made this statement to counter the widespread fears of automation-driven unemployment, arguing instead that AI will lower the barrier to complex tasks and empower a much larger portion of the population to participate in the workforce.
Huang has argued that the returns of deploying AI in workplaces are already visible. By investing $10,000 in AI tools per $100,000 employee, companies can achieve 2–3x productivity gains, and that this is already playing out across Nvidia’s engineering and chip design teams.
However, various industry reports say that outside Nvidia, the results are quite the opposite. Most companies experimenting with AI are seeing small, gradual improvements, and not the dramatic leaps that justify billion-dollar infrastructure bets. According to MIT’s GenAI Divide report this July, a staggering 95 percent of US firms investing in in-house AI projects have seen zero measurable returns on investments (ROI). These companies collectively spent between $35–40 billion, yet only 5 percent managed to extract any real business value. A Forbes CIO survey in January 2025 echoes this finding, and shows that 75 percent of businesses haven’t seen any meaningful returns from their AI deployments.
Alongside the doubts surrounding the tangible benefits of deploying AI at workplaces, there’s growing scrutiny around how these deals are structured. OpenAI is expected to spend much of Nvidia’s investment on buying Nvidia’s own chips, thus raising concerns about circular financing, where companies invest in their customers’ companies to drive demand for their own products. Some analysts have compared this to the “round-tripping” seen during the dotcom bubble.
For instance, in an investor note released shortly after Nvidia announced its investment in OpenAI, Stacy Rasgon, senior analyst at Bernstein Research, wrote that the deal “will clearly fuel ‘circular’ concerns,” adding that its size “appears to dwarf all the others”, likely intensifying worries about the rationale behind such investments. Leading British tech investor James Anderson voiced his concerns to Financial Times, saying that soaring valuations of AI firms such as OpenAI and Anthropic are “disconcerting”, and compared it to the concept of “vendor financing” that was prevalent during the dotcom bubble.
First Published: Oct 01, 2025, 17:41
Subscribe Now