 (4).png)
1 week ago
2
ARTICLE AD BOX
Summary
Valued at $20 billion, the deal is one of the largest in the sector and has reignited concerns about a possible AI investment bubble and ‘circular financing’ by companies in the space.
On Christmas Eve, Nvidia announced a deal to license technology, acquire assets, and hire key team members from AI chip startup Groq. Valued at $20 billion, the deal is one of the largest in the sector and has reignited concerns about a possible AI investment bubble and ‘circular financing’ by companies in the space.
However, there are positive signs as well. Business adoption of AI is accelerating, early benefits are being reported, and models continue to improve. Governments worldwide are increasing their focus on the sector, providing incentives and creating regulations. How the industry adapts to these tightening rules will be critical in the coming years.
Circular deals
Companies have committed billions to AI infrastructure through deals that increasingly blur the line between investment and vendor financing. In September 2025, Nvidia announced plans to deploy up to $100 billion in AI data centre capacity for OpenAI, contingent on OpenAI buying millions of Nvidia chips as the systems are built. And in October, OpenAI entered into a deal with AMD that grants OpenAI warrants for up to a 10% stake in the chip company, vesting as it deploys AMD processors.
Such deals have a sparked debate about circular financing, in which invested capital flows back to the investor through obligated purchases. Analysts have drawn parallels to vendor financing schemes that historically inflated demand in the telecom and internet sectors, raising questions about the sustainability of investments in AI infrastructure.
OpenAI’s finances underscore these tensions. Valued around $500 billion, the company generated $4.3 billion in revenue during the first half of the year while burning $2.5 billion in cash and recording operating losses in excess of $7 billion. AI-related company valuations have surged by more than $19 trillion since late 2022, running at or above the upper bound of plausible long-term economic returns, according to Goldman Sachs estimates.
Overbuild risks
Tech majors have sharply increased their AI infrastructure spending as they expect the demand for compute capacity to grow. Microsoft, Amazon, Meta and Alphabet and Oracle spent a combined $399 billion on this in 2025. Bank of America expects it to touch $620 billion by 2028.
The buildout has fueled dealmaking. Data-centre transactions totalled nearly $61 billion through November 2025, according to S&P Global Market Intelligence, a record. Goldman Sachs estimates that global AI infrastructure spending could rise to $3-4 trillion by 2030.
This pace has raised sustainability concerns. Capital expenditure now acounts for half or more of operating cash flow for several hyperscalers, according to CreditSights estimates. In February 2025, Microsoft CEO Satya Nadella warned that “there will be an overbuild” of AI capacity.
Even so, there are expectations that demand will match supply. Apollo Global Management has noted that AI investment remains below 1% of US GDP, well under past technology-boom peaks, suggesting the expansion may still have room to run.
Value capture
A reason for this optimism is that the corporate adoption of AI has accelerated sharply. McKinsey's 2025 Global Survey found that 88% of organisations now use AI in at least one business function, up from 78% in 2024 and roughly 50% before generative AI's emergence. Deloitte, another consultancy, found that 79% of CEOs were implementing generative AI for innovation.
Larger organisations are adopting AI faster. “Nearly half of respondents from companies with more than $5 billion in revenue have reached the scaling phase, compared with 29% of those with less than $100 million in revenues,” McKinsey said in its November report.
Some have started reporting tangible benefits. Software engineering, manufacturing and IT led on cost benefits, according to McKinsey. However, only 6% of organisations generate substantial earnings before interest and taxes (Ebit) from AI. Such gains depend on systematic workflow redesign rather than isolated pilots. This pattern of widespread adoption but concentrated benefits suggests the market is still in its early stages. But the shift has happened.
Rapid releases
The past year saw intense competition among AI leaders, leading to improvements in models, which could spark wider adoption by businesses. Major AI companies including OpenAI, Google, Anthropic and xAI released powerful new models that doubled performance on complex tasks compared to the previous year. There was a particularly intense burst of releases in November and December. These systems achieved impressive performance in coding and mathematics, some even earning "gold medal" scores on advanced math competitions.
For example, Google's Gemini 3, launched on 18 November, topped several public leaderboards including LMSYS Arena, a community-driven platform where users compare and vote for responses from different AI models. OpenAI responded by declaring an internal "code red" and accelerated the release of its latest version.
The AI landscape is dominated by a small group of companies, led by Google with 187 notable models from 2014-24, followed by Meta (82), Microsoft (39), and OpenAI (36), according to the Stanford AI Index. This could change in the coming years. In 2025, Chinese developers matched or exceeded their Western counterparts in open-source development.
Regulatory surge
Global competition is expected to intensify, with increased support from national governments. Saudi Arabia, France and China have committed billions of dollars to supporting AI in their respective regions. “Nations are seeking 'sovereignty' for the same reason they have domestic utilities, manufacturing, borders, armies, and currencies: to control their own destiny,” Stanford's 2025 AI Index Report said.
Regulatory activity has surged in parallel, suggesting the sector is maturing beyond purely market-driven growth.
US federal agencies introduced 59 AI-related regulations in 2024, more than double the prior year's total, according to the Stanford report. Legislative mentions of AI across 75 countries increased 21.3 % to 1,889. US states enacted 131 AI-related laws in 2024, up from 49 in 2023.
The EU AI Act came into force in August 2024, with phased obligations beginning in 2025. International bodies including the OECD, UN, and African Union released governance frameworks emphasising transparency and trustworthiness.
While businesses recognise AI risks around bias and privacy, implementation of mitigation tools lags significantly, according to the Stanford report. How the industry keeps pace will be critical in the coming years.
www.howindialives.com is a database and search engine for public data.
English (US) ·