Rise of the Machines: AI Agents Now Burn 5x More Tokens Than Humans — and the Gap Is Widening Fast
AI agents are consuming tokens at a pace that dwarfs human usage, and the companies supplying the hardware behind every request are seeing that demand translate into numbers that few investors expected to see this soon.
This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
One Number: $89.02 Billion From the Machines That Make Tokens
NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) generated $89.02 billion in Data Center revenue in a single quarter. That figure covers the second quarter of fiscal 2027, reported on August 26, 2026, and it rose 117% year over year. Total company revenue for the quarter was $96.22 billion, so the Data Center segment produced almost all of it. That segment holds the GPUs, CPUs and networking gear that AI models run on to produce tokens.
Token demand is also changing shape. OpenRouter data shows agentic token usage rose from roughly 0.51 trillion tokens in February to about 7.3 trillion by August, a fourteenfold increase. Human usage grew only about 2.8 times over the same stretch, to approximately 1.4 trillion tokens. AI agents now use roughly five times more tokens than people do. NVIDIA’s $89.02 billion is what that shift looks like on an income statement.
What $89.02 Billion Reveals About Agent-Driven Demand
The growth is accelerating. Data Center revenue rose 18% quarter over quarter, and total revenue growth sped up for the fourth consecutive quarter. Hyperscale customers contributed $49 billion, up 13% sequentially. The group that covers neoclouds, enterprises, sovereign customers and industrial buyers grew faster, reaching $40 billion, up 25% sequentially and 138% year over year. Demand is spreading beyond a handful of cloud giants.
Agents are the main reason. According to the OpenRouter data, agents use roughly fifteen times more tokens per request than people. One employee action can set off dozens of model calls, tool runs and retries. Chief Executive Jensen Huang described where this is heading on the earnings call: “When the world goes to agentic, fully agentic systems, you’re going to have agents running all the time, working with other agents running all the time.”
NVIDIA is making more money on each dollar of sales as it grows. Non-GAAP gross margin expanded to 75.0% from 72.5% a year earlier, and net income reached $59.69 billion, up 125.9%. Huang summed it up: “AI has reached its inflection point. It’s doing useful work. Its tokens are productive and profitable. Now, compute is revenue.”
Shares Have Climbed Since the August Earnings Report
Shares were priced at $213.67 when the results came out on August 26, 2026, and closed at $233.95 on October 2, 2026. The stock is up 25.74% so far this year and gained 4.37% over the past month.
Bull Case: Demand Outruns NVIDIA’s Supply
NVIDIA can’t build enough to meet demand. Management said “NVIDIA Compute is fully utilized across every cloud we serve” and that the company has supply for about 70% of expected demand. NVIDIA expects fiscal 2028 revenue to grow about 70%, and it says supply is what limits that number. Customers are the ones asking for more.
Each new chip generation also brings in more revenue per unit of power. NVIDIA puts its revenue opportunity per gigawatt at roughly $18 billion for Hopper, about $25 billion for Grace Blackwell and about $40 billion for Vera Rubin. Supply and purchase obligations have climbed to $279.00 billion, mostly to buy memory for Vera Rubin production. That shows how much NVIDIA is investing to the ramp ahead.
Shareholders are getting paid along the way. NVIDIA returned about $26.0 billion in the quarter and still has about $99.0 billion left on its buyback authorization. Analysts keep raising estimates: over the past 30 days, fiscal 2028 EPS forecasts got 42 upward revisions and 0 downward ones.
Risks remain: about 85% of agentic tokens come from cached prompts charged at lower rates. Gross margin is expected to bottom at 71% to 72% in the fourth quarter as memory costs rise. Days sales outstanding, a measure of how long customers take to pay, extended to 60 days from 45. Online sentiment has also cooled, with the overall sentiment score falling to 37.17. Even so, cached tokens still need servers, memory and power to run. The supply shortages NVIDIA flagged in memory, power, cooling and networking also bring in business for Micron Technology (NASDAQ:MU), Vertiv (NYSE:VRT), Eaton (NYSE:ETN) and Arista Networks (NYSE:ANET). We profiled seven of these picks-and-shovels suppliers, from power to cooling to networking, in a free report you can grab here.
Bottom Line: Third-Quarter Guidance Is the Next Checkpoint
The $89.02 billion figure shows token demand turning into revenue, and agents are adding to that demand faster than people are. NVIDIA’s own guidance gives the next checkpoint: third-quarter revenue of $108.0 billion, plus or minus 2%, which excludes China data center compute. Vera Rubin is expected to make up about 20% of data center revenue in Q3. Long-term holders should note that AI agents now run around the clock, and each request they make adds to the demand for NVIDIA’s hardware.
Contact [email protected] for any questions or corrections.
After two decades of patrolling the dark corners of suburbia as a police officer, Rich Duprey hung up his badge and gun to begin writing full time about stocks and investing. For the past 20 years, he’s been cruising the markets looking for companies to lock up as long-term holdings in a portfolio while writing extensively on the broad sectors of consumer goods, technology, and industrials. Because his experience isn’t from the typical financial analyst track, Rich is able to break down complex topics into understandable and useful action points for the average investor. His writings have appeared on The Motley Fool, InvestorPlace, Yahoo! Finance, Money Morning, and, of course, 24/7 Wall St. He has been featured in both U.S. and international publications, including MarketWatch, Financial Times, Forbes, Fast Company, and USA Today.
