Boom or bubble: How long can the AI investment craze last?

The wave of investment in artificial intelligence shows no signs of slowing down. Recently, Nvidia, a leading AI chip manufacturer, announced a staggering commitment of $100 billion to assist OpenAI, a key player in generative AI, in building extensive data centers. This raises the question: how can such colossal sums be justified when current returns on these investments appear limited? Worldwide spending on AI is set to skyrocket, with projections estimating it will reach around $1.5 trillion by 2025 and exceed $2 trillion by 2026, representing nearly 2% of the global GDP, according to Gartner, a US research firm. Despite the fact that immediate returns are lagging behind the level of investment, the momentum behind the AI revolution seems unstoppable. Denis Barrier, the head of investment fund Cathay Innovation, asserts that investors view AI as a groundbreaking technology comparable to the advent of electricity. In Silicon Valley, the prevailing attitude is one of seizing opportunities rather than dwelling on potential risks. Geopolitical tensions are contributing to this investment frenzy, particularly as companies strive to establish massive data centers that house thousands of costly chips, necessitating substantial energy and cooling resources. From 2013 to 2024, private investment in AI within the United States totaled approximately $470 billion, with almost a quarter of that occurring in the last year alone. China follows with $119 billion in investments, as reported by Stanford University. OpenAI has emerged as the primary beneficiary of this influx of funding. In March 2025, it raised around $40 billion, boosting its valuation to about $300 billion. Notably, OpenAI has now surpassed SpaceX as the most valuable company in the world, hitting a market cap of $500 billion after a deal allowing employees to sell a limited number of shares. The company, led by CEO Sam Altman, is at the heart of an expansive AI investment initiative, overseeing the Stargate project, which has secured $400 billion of a planned $500 billion for data centers in Texas. However, Nvidia has faced scrutiny for engaging in what some describe as 'circular funding'—investing in startups that subsequently use those funds to purchase Nvidia chips. Critics warn that this could lead to a bubble. According to Stacy Rasgon, an analyst at Bernstein Research, the recent OpenAI deal may heighten these concerns. Reports indicate that OpenAI generated roughly $4.3 billion in revenue in the first half of 2025, suggesting that unlike tech giants like Meta or Google, which have significant cash reserves, OpenAI and its rivals, such as Anthropic and Mistral, must innovate in securing funding to stay afloat. Despite the potential for significant revenue growth, challenges loom ahead. Bain & Company estimates that global investments in data centers to support AI could reach $500 billion annually through 2030, necessitating $2 trillion in yearly revenues to make these expenses sustainable. Even under optimistic scenarios, the AI sector could face an $800 billion deficit. OpenAI alone anticipates spending over $100 billion by 2029, indicating that profitability may still be far off. The energy demands of AI are equally daunting, with projections suggesting that its global computing footprint might hit 200 gigawatts by 2030—equivalent to Brazil's annual electricity consumption, with half of that expected to be in the United States. Nevertheless, many analysts remain hopeful. Dan Ives from Wedbush Securities likens the current state of the AI sector to the early days of the internet boom, predicting that while there will be losses and failures, as seen in the past, the AI industry will ultimately endure and thrive.