Nvidia CEO Warns of AI Investment Returns Amid Job Cuts and Token Budgeting
# Companies Switched People for Tokens, But Gains Are Lacking
**Nvidia CEO Raises Concerns Over Return on AI Investments**
Jensen Huang, the CEO of Nvidia, has set a benchmark for evaluating the value of engineers based on their annual AI token usage. Speaking on the All-In Podcast following the conclusion of GTC 2026, Huang expressed alarm if an engineer with a $500,000 salary consumed less than $250,000 in tokens, which amounts to 50% of their pay. He disclosed that Nvidia is targeting an annual token expenditure of $2 billion for its engineering team.
This provocative stance reflects a broader trend in corporate budgeting where funds once allocated for personnel are increasingly redirected towards token investments. However, the pressing question across the industry is whether this strategic shift is yielding the desired results. Companies that have made this transition first are showing that the anticipated benefits often do not materialise.
## Where the Budget Went
The reallocation of funds is widely acknowledged. The four leading hyperscalers are expected to have a combined capital expenditure of around $700 billion in 2026, nearly double from the previous year. In parallel, Gartner forecasts that spending on AI agent software will surge to $207 billion, representing a remarkable 139% increase. On the flip side, research from Challenger, Gray & Christmas indicates that for the fourth consecutive month, AI has been cited as the primary reason for job losses in the U.S., with the tech sector accounting for 31% of layoffs in the first half of the year.
An internal memo from Meta, obtained by Reuters, revealed that it cut 8,000 jobs in May to balance its substantial investments, despite seeing a 33% revenue increase that quarter. Oracle's reports show a reduction of 21,000 employees as savings support its data center development. These firms are not in financial distress; instead, they are strategically reallocating resources.
Andy Challenger summarised the data succinctly: “Companies are shifting budgets toward AI investments at the expense of jobs.”
## What the Investment Yielded
The results of these financial strategies are less impressive. A Gartner survey of 350 executives from companies with revenues exceeding $1 billion, all implementing AI technologies, found that about 80% had reduced their workforce. However, a direct correlation between these layoffs and improved returns was not evident. Analyst Helen Poitevin concluded, “Workforce reductions may create budget room, but they do not create return.”
Furthermore, organisations that successfully improved returns did so by using AI to enhance their workforce rather than replace employees.
## Excess Budget for Token Use
Uber's experience highlights the challenges with token investments. The company provided 5,000 engineers with AI coding tools in December but exhausted its full AI budget by April. COO Andrew Macdonald noted that while 70% of the code created was AI-generated, it lacked a direct connection to customer experiences, stating, “That link is not there yet.” Now, Uber has limited its engineers to a $1,500 monthly cap on AI expenditures.
Similarly, Walmart faced a surge in demand for its internal AI assistant, leading to rationing of token use, as reported by Bloomberg. The stark contrast is evident: when token usage exceeds budgetary limits, it gets restricted; but when personnel costs go over budget, it results in layoffs.
## A Public Admission
Klarna, the well-known fintech company, provides a clear example of these dynamics. The company replaced approximately 700 customer service roles with an OpenAI-driven assistant, halting human hiring for over a year, even presenting its AI-centric model to potential investors. However, this approach backfired as customer satisfaction declined and complaints increased. CEO Sebastian Siemiatkowski publicly acknowledged, “We focused too much on efficiency and cost. The result was lower quality, and that’s not sustainable.” Klarna is now reinstating human roles, positioning them as vital to the company’s future.
Gartner predicts that by 2027, many companies that reduced customer service staffing in favour of AI will rehire, often in redefined roles. Their survey of 321 customer service leaders indicated that only 20% had genuinely decreased staff due to AI, suggesting that many cuts were simply cost-cutting measures disguised as AI-related.
OpenAI's Sam Altman has recognised the issue, admitting instances of “AI washing” surrounding layoffs, while venture capitalist Marc Andreessen has referred to AI as the “silver bullet excuse.” Thus, while the narrative around worker displacement holds some truth, the underlying budget shift is undeniable and problematic.
## Consequences of This Experiment
The consequences primarily affect those least equipped to handle them. The Stanford HAI’s 2026 AI Index revealed a nearly 20% drop in employment among software developers aged 22 to 25 since 2024, while older age groups have maintained steady growth. Companies are effectively eliminating entry-level positions, expecting senior engineers—who manage token allocations—to remain employed in the future.
The global picture is even more dire. Huang's scenario presumes a starting salary of $500,000 for engineers, a figure that encompasses merely 2% to 5% of American software workers, with far fewer in other regions. If the token budget is applied to engineers in places like Kuala Lumpur, Manila, or Jakarta, the token costs could eclipse actual salaries.
What Klarna experienced, resulting in 700 job losses and damage to its brand, aligns with Gartner’s aggregate data: companies that prioritise investments in personnel who utilise AI, rather than AI that displaces staff, are poised for greater returns. CFOs who are now placing caps on token budgets, having exhausted them within a quarter, are beginning to rediscover a lesson the industry had overlooked for two years: talent was never the barrier to success.
