Only a few years back, AI was an emerging idea producing strange, low-quality images and videos that showed up in social media streams. Now, it appears to be everywhere. Fresh models emerge nearly monthly. Hollywood is incorporating AI in certain areas. And although it hasn’t yet improved office productivity, AI has probably already entered your work environment. This vast growth demands huge infrastructure spending. Nvidia chief executive Jensen Huang stated his firm anticipates supplying those fundamental components on a grand scale.
In his Monday keynote speech at Nvidia’s GTC event in San Jose, Huang revealed the firm has doubled its demand projection for the coming year. “I anticipate at least $1 trillion through 2027,” he stated. “Actually, we’ll fall short. I’m convinced computing needs will far exceed that amount.”
He’s already gearing up for that scenario with an unconventional perk to draw elite talent and extract greater computing capacity from staff: providing engineers with AI tokens valued at close to 50% of their compensation.
The AI surge is driving infrastructure spending to unprecedented levels. Technology firms are pouring a massive $700 billion into constructing data centers, an amount comparable to the GDP of advanced nations like Sweden, and exceeding twice the inflation-adjusted expense of the Apollo moon missions. Nvidia serves as a vital vendor in this expansion, supplying the chips that drive AI facilities. The $1 trillion demand estimate further demonstrates this construction is accelerating without slowdown, even as rivals such as Advanced Micro Devices (AMD) find it difficult to narrow the lead. This occurs despite growing concerns about an AI bubble, as noted by executives like Microsoft CEO Satya Nadella and “Big Short” investor Michael Burry.
Huang issued this forecast while also asserting that AI agents might soon manage global operations, along with declarations regarding orbital computing intended to deploy AI in space, an idea Elon Musk has highlighted as a possible answer to the power requirements of growing data centers.
“We’re entirely rebooting and initiating the biggest infrastructure expansion in human history,” Huang remarked. “Most global industries constructing AI factories, semiconductor facilities, and computer manufacturing plants are present here today.”
The firm’s latest financial results have lent weight to Huang’s assertions. In the previous month, Nvidia reported $215.9 billion in revenue for fiscal 2026, a 65% increase from the prior year, marking its highest annual performance ever. Data center income alone climbed 75% year-over-year to $62.3 billion.
AI Tokens: The Future of Compensation?
While corporate executives seek to leverage AI to enhance employee efficiency, Huang provided insight into Nvidia’s strategy for implementing that goal: compensating engineers with tokens—the medium of exchange in AI—to magnify their productivity.
“I can easily envision a time when every engineer at our company will require an annual token allocation,” he said. “They’ll earn a few hundred thousand annually as base salary. I’ll likely provide them with half that amount again in tokens so they can become ten times more effective.”
Tokens represent the fundamental data or word units that AI systems utilize to interpret language and identify patterns, rendering them essential for future AI implementation. According to AI firm OpenAI, one token corresponds to roughly four characters, while a brief one- or two-sentence query needs approximately 30 tokens. The word “Magazine,” for instance, might be split into five tokens: “For” “tune” “Mag” “az” “ine.”
Under the allocation levels Huang outlined, engineers would receive billions of tokens each year, releasing a flood of computational capacity. In Huang’s vision, tokens would serve as an additional job benefit for his company’s engineers, equipping them with the resources necessary to perform intensive corporate research.
The Nvidia chief executive predicted other technology companies will rapidly adopt similar practices and employ tokens as a hiring mechanism to lure premier sector talent.
“It’s already become a hiring incentive in Silicon Valley: what token allowance accompanies my position,” he stated. “The rationale is obvious because any engineer with token access will achieve greater productivity.”