Bezos, Jensen Huang and Masayoshi Son Clash Over AI Boom

Three technology leaders publicly debated the scale, costs and risks of artificial intelligence as investors directed roughly $380 billion into AI-linked companies year-to-date. The remarks came as major firms tightened internal controls on AI spending.

Nvidia CEO Jensen Huang dismissed claims that AI will reduce jobs, saying “People are talking about AI decreasing jobs, it’s complete nonsense.” SoftBank founder Masayoshi Son estimated the current cycle could far exceed the 1990s internet boom, saying “I think this is like more than 10x, probably 50x bigger than dotcom.” Jeff Bezos described the moment as an industrial bubble and noted that speculative excess can leave behind useful infrastructure when weaker projects fail.

Investors routed capital to AI through three main channels this year: roughly $140 billion in investment-grade bonds, nearly $220 billion in venture capital, and about $21 billion in high-yield credit. Those totals represent about 49% of investment-grade issuance, 87% of venture funding and roughly 64% of the tracked capital flows across the three channels.

SoftBank announced a €75 billion commitment to build about 5 gigawatts of AI data-center capacity in France during a visit with France’s president. The pledge adds to a wave of data-center and chip investment as companies increase computing capacity to train and run advanced models.

Companies have moved to curb running costs. An Amazon engineering executive asked staff to stop using AI for trivial tasks after the company reportedly spent roughly $500 million on tokens in a single month. Other firms, including Uber, Salesforce, Meta and Microsoft, circulated internal cautions to reduce unnecessary AI queries. Analysts say hyperscale cloud providers’ free cash flow has fallen to near a decade low.

Estimates of per-query costs vary. VanEck strategist Matthew Sigel said a leading model can summarize a 500-page book for about $2.50. Other research indicates AI infrastructure can be substantially more expensive than legacy systems, creating a cost gap investors are watching.

A separate projection by commentator Will Sommer estimated hyperscale cloud providers would need about $7 trillion in revenue over the next three years to achieve a 7% return on invested capital, a figure cited by market participants when assessing financing levels and capacity build-outs.

Companies and investors said upcoming earnings reports will show whether AI investments are translating into revenue and margins that match the scale of recent capital spending.