Key Takeaways
Nvidia vice chairman Bryan Catanzaro says that for his crew, AI compute now prices greater than the workers utilizing it, making AI costlier than human labor.
A 2024 MIT examine finds AI automation is economically viable in solely about 23% of jobs, with people nonetheless cheaper within the remaining 77%.
Regardless of unclear productiveness positive factors and excessive prices, huge tech firms have dedicated round $740 billion to AI-related bills this yr, a 69% soar from 2025.
A key Nvidia government says that AI isn’t decreasing labor prices — proper now, it’s truly costlier than the human staff that firms have already got.
“For my crew, the price of compute is much past the prices of the workers,” Bryan Catanzaro, vice chairman of utilized deep studying at Nvidia, just lately instructed Axios.
A 2024 MIT examine helps this view. Researchers checked out what it might take for AI programs to match human efficiency throughout completely different jobs and located that automation made monetary sense in simply 23% of roles that rely closely on visible duties. Within the different 77% of instances, holding human staff was nonetheless the cheaper choice.
There are additionally examples of AI making pricey errors. In a single case, an engineer stated an AI software worn out his database and community.
Firms are investing in AI
Regardless of the drawbacks of AI, huge tech firms are nonetheless investing closely in it. In response to Morgan Stanley, tech corporations have already dedicated about $740 billion to AI-related spending this yr, a 69% improve from 2025. Over simply the previous yr, charges for AI software program have additionally gone up sharply, rising by 20% to 37%, in accordance with spending administration firm Tropic.
AI spending is rising rapidly. Based mostly on McKinsey estimates, it may attain $5.2 trillion by 2033, together with about $1.6 trillion for information facilities and $3.3 trillion for IT {hardware}.
The dimensions of that funding is so massive that some firms are actually reconsidering how they allocate their budgets. For instance, Uber’s chief know-how officer, Praveen Neppalli Naga, instructed The Data earlier this month that the corporate’s shift towards AI coding instruments is driving up prices. “I’m again to the drafting board as a result of the funds I assumed I would want is blown away already,” he stated.
Tech layoffs are on the rise
As firms improve AI spending, layoffs throughout the tech trade have been rising. Information from layoff tracker Layoffs.fyi reveals that greater than 92,000 tech staff have already misplaced their jobs this yr, spanning almost 100 firms. That tempo is far sooner than final yr, when roughly 120,000 layoffs occurred over the complete yr.
Keith Lee, an AI and finance professor on the Swiss Institute of Synthetic Intelligence’s Gordon College of Enterprise, instructed Fortune that firms are spending enormous quantities on AI, regardless that human staff are presently cheaper for a lot of duties. There’s a niche between what makes monetary sense on paper and what firms are literally doing. “What we’re seeing is a short-term mismatch,” Lee instructed the outlet.
AI could also be costlier than human staff proper now, however that might change. Lee says the economics will shift as the price of operating AI fashions drops and infrastructure improves. He added that AI will solely turn out to be really cost-effective if it proves dependable and desires much less human supervision.
“It’s not nearly AI changing into cheaper than people,” Lee instructed Fortune. “It’s about changing into each cheaper and extra predictable at scale.”
Key Takeaways
Nvidia vice chairman Bryan Catanzaro says that for his crew, AI compute now prices greater than the workers utilizing it, making AI costlier than human labor.
A 2024 MIT examine finds AI automation is economically viable in solely about 23% of jobs, with people nonetheless cheaper within the remaining 77%.
Regardless of unclear productiveness positive factors and excessive prices, huge tech firms have dedicated round $740 billion to AI-related bills this yr, a 69% soar from 2025.
A key Nvidia government says that AI isn’t decreasing labor prices — proper now, it’s truly costlier than the human staff that firms have already got.
“For my crew, the price of compute is much past the prices of the workers,” Bryan Catanzaro, vice chairman of utilized deep studying at Nvidia, just lately instructed Axios.
A 2024 MIT examine helps this view. Researchers checked out what it might take for AI programs to match human efficiency throughout completely different jobs and located that automation made monetary sense in simply 23% of roles that rely closely on visible duties. Within the different 77% of instances, holding human staff was nonetheless the cheaper choice.
