Finals season looks different this year. Across college campuses, students are slogging their way through exams with all-nighters and lots of caffeine, just as they always have. But they’re also getting more help from AI than ever before. Through the end of May, OpenAI is offering students two months of free access to ChatGPT Plus, which normally costs $20 a month. It’s a compelling deal for students who want help cramming—or cheating—their way through finals: Rather than firing up the free version of ChatGPT to outsource essay writing or work through a practice chemistry exam, students are now able to access the company’s most advanced models, as well as its “deep research” tool, which can quickly synthesize hundreds of digital sources into analytical reports.

The OpenAI deal is just one of many such AI promotions going around campuses. In recent months, Anthropic, xAI, Google, and Perplexity have also offered students free or significantly discounted versions of their paid chatbots. Some of the campaigns aren’t exactly subtle: “Good luck with finals,” an xAI employee recently wrote alongside details about the company’s deal. Even before the current wave of promotions, college students had established themselves as AI’s power users. “More than any other use case, more than any other kind of user, college-aged young adults in the US are embracing ChatGPT,” the vice president of education at OpenAI noted in a February report. Gen Z is using the technology to help with more than schoolwork; some people are integrating AI into their lives in more fundamental ways: creating personalized workout plans, generating grocery lists, and asking chatbots for romantic advice.

AI companies’ giveaways are helping further woo these young users, who are unlikely to shell out hundreds of dollars a year to test out the most advanced AI products. Maybe all of this sounds familiar. It’s reminiscent of the 2010s, when a generation of start-ups fought to win users over by offering cheap access to their services. These companies especially targeted young, well-to-do, urban Millennials. For suspiciously low prices, you could start your day with pilates booked via ClassPass, order lunch with DoorDash, and Lyft to meet your friend for happy hour across town. (On Uber, for instance, prices nearly doubled from 2018 to 2021, according to one analysis). These companies, alongside countless others, created what came to be known as the “Millennial lifestyle subsidy.” Now something similar is playing out with AI. Call it the Gen Z lifestyle subsidy. Instead of cheap Ubers and subsidized pizza delivery, today’s college students get free SuperGrok.

AI companies are going to great lengths to chase students. Anthropic, for example, recently started a “campus ambassadors” program to help boost interest; an early promotion offered students at select schools a year’s worth of access to a premium version of Claude, Anthropic’s AI assistant, for only $1 a month. One ambassador, Josefina Albert, a current senior at the University of Washington, told me that she shared the deal with her classmates, and even reached out to professors to see if they might be willing to promote the offer in their classes. “Most were pretty hesitant,” she told me, “which is understandable.”

The current discounts come at a cost. There are roughly 20 million postsecondary students in the U.S. Say just 1 percent of them take advantage of free ChatGPT Plus for the next two months. The start-up would effectively be giving a handout to students that is worth some $8 million. In Silicon Valley, $8 million is a rounding error. But many students are likely taking advantage of multiple such deals all at once. And, more to the point, AI companies are footing the bill for more than just college students. All of the major AI companies offer free versions of their products despite the fact that the technology itself isn’t free. Every time you type a message into a chatbot, someone somewhere is paying for the cost of processing and generating a response. These costs add up: OpenAI has more than half a billion weekly users, and only a fraction of them are paid subscribers. Just last week, Sam Altman, the start-up’s CEO, suggested that his company spends tens of millions of dollars processing “please” and “thank you” messages from users. Tack on the cost of training these models, which could be as much as $1 billion for the most advanced versions, and the price tag becomes even more substantial. (The Atlantic recently entered into a corporate partnership with OpenAI.)

These costs matter because, despite AI start-ups’ enormous valuations (OpenAI was just valued at $300 billion), they are wildly unprofitable. In January, Altman said that OpenAI was actually losing money on its $200-a-month “Pro” subscription. This year, the company is reportedly projected to burn nearly $7 billion; in a few years, that number could grow to as much as $20 billion. Normally, losing so much money is not a good business model. But OpenAI and its competitors are able to focus on acquiring new users because they have raised unprecedented sums from investors. As my colleague Matteo Wong explained last summer, Silicon Valley has undertaken a trillion-dollar leap of faith, on track to spend more on AI than what NASA spent on the Apollo space missions, with the hope that eventually the investments will pay off.

The Millennial lifestyle subsidy was also fueled by extreme amounts of cash. Ride-hailing businesses such as Uber and Lyft scooped up customers even as they famously bled money for years. At one point in 2015, Uber was offering carpool rides anywhere in San Francisco for just $5 while simultaneously burning $1 million a week. At times, the economics were shockingly flimsy. In 2019, the owner of a Kansas-based pizzeria noticed that his restaurant had been added to DoorDash without his doing. Stranger still, a pizza he sold for $24 was priced at $16 on DoorDash, yet the company was paying him the full price. In its quest for growth, the food-delivery start-up had reportedly scraped his restaurant’s menu, slapped it on their app, and was offering his pie at heavy discount. (Naturally, the pizzeria owner started ordering his own pizzas through DoorDash—at a profit.)

These deals didn’t last forever, and neither can free AI. The Millennial lifestyle subsidy eventually came crashing down as the cheap money dried up. Investors that had for so long allowed these start-ups to offer services at artificially deflated prices wanted returns. So companies were forced to raise prices, and not all of them survived.

If they want to succeed, AI companies will also eventually have to deliver profits to their investors. Over time, the underlying technology will get cheaper: Despite companies’ growing bills, technical improvements are already increasing efficiency and driving down certain expenses. Start-ups could also raise revenue through ultra-premium enterprise offerings. OpenAI is reportedly considering selling “PhD-level research agents” at $20,000 a month. But it’s unlikely that companies such as OpenAI will allow hundreds of millions of free users to coast along indefinitely. Perhaps that’s why the start-up is currently working on both search and social media; Silicon Valley has spent the past two decades essentially perfecting the business models for both.

Today’s giveaways put OpenAI and companies like it only further in the red for now, but maybe not in the long run. After all, Millennials became accustomed to Uber and Lyft, and have stuck with ride-hailing apps even as prices have increased since the start of the pandemic. As students learn to write essays and program computers with the help of AI, they are becoming dependent on the technology. If AI companies can hook young people on their tools now, they may be able to rely on these users to pay up in the future.

Some young people are already hooked. In OpenAI’s recent report on college students’ ChatGPT adoption, the most popular category of non-education or career-related usage was “relationship advice.” In conversations with several younger users, I heard about people who are using AI for color-matching cosmetics, generating customized grocery lists based on budget and dietary preferences, creating personalized audio meditations and half-marathon training routines, and seeking advice on their plant care. When I spoke with Jaidyn-Marie Gambrell, a 22-year-old based in Atlanta, she was in the parking lot at McDonald’s and had just consulted ChatGPT on her order. “I went on ChatGPT and I’m like, ‘Hey girl,’” she said. “‘Do you think it’d be smart for me to get a McChicken?’” The chatbot, which she has programmed to remember her dietary and fitness goals, advised against it. But if she really wanted a sandwich, ChatGPT suggested, she should order the McChicken with no mayo, extra lettuce, tomatoes, and no fries. So that’s what she got.

The Gen Z lifestyle subsidy isn’t entirely like its Millennial predecessor. Uber was appealing because using an app to instantly summon a car is much easier than chasing down a cab. Ride-hailing apps were destructive for the taxi business, but for most users, they were just convenient. Today’s chatbots also sell convenience by expediting essay writing and meal planning, but the technology’s impact could be even more destabilizing. College students currently signing up for free ChatGPT Plus ahead of finals season might be taking exams intended to prepare them for jobs that the very same AI companies suggest will soon evaporate. Even the most active young users I spoke with had mixed feelings about the technology.  Some people “are skating through college because of ChatGPT,” Gambrell told me. “That level of convenience, I think it can be abused.” When companies offer handouts, people tend to take them. Eventually, though, someone has to pay up.