Chegg closed at $0.96 a share on July 24, 2026, a company once valued near $14 billion now worth just over $100 million. Stack Overflow’s monthly question count, which topped 200,000 in 2014, has fallen to roughly 300. Getty Images and Shutterstock’s merger, built as a joint defense against exactly this kind of disruption, died on July 7 when the two companies let the deal lapse rather than accept a British regulator’s terms.
Art Zeile, chief executive of DHI Group, the parent company of tech job boards Dice and ClearanceJobs, has become one of the clearest voices explaining why. His framework for sorting winners from losers holds up against the data. It also happens to point straight at his own company’s numbers.
A Dollar Stock Where $14 Billion Used to Sit
Chegg was the first public company to say the quiet part out loud. On a May 2023 earnings call, then-CEO Dan Rosensweig told investors ChatGPT was having “an impact on our new customer growth rate,” and the stock lost half its value the next morning.
Three years later the damage looks total. Chegg cut its workforce twice, 248 employees in May 2025 and then roughly 388 more, about 45% of what remained, as it confronted what the company called a steep drop in traffic and revenue tied to generative AI and changing search habits. Its most recent quarterly report showed net revenue down 49% year over year to $72.7 million and a net loss of $32.8 million. The company says it is still operating and now pushing into skills training and language learning to find ground that holds.
- $0.96: Chegg’s closing share price on July 24, 2026
- 99%: approximate decline from its peak market value near $14 billion
- 636: employees cut across two rounds of layoffs since May 2025
- 49%: year-over-year drop in net revenue in its latest quarterly report
Chegg’s own restructuring language named a specific culprit: reduced traffic from Google to content publishers, the same complaint behind publishers challenging Google’s AI Overviews before EU regulators. Different companies, identical mechanism. When the answer arrives inside the search result, the site that used to sell the answer stops getting the click.
Two Axes That Predict Who Survives
Zeile’s framework maps information businesses on a simple grid. One axis runs from answers at the top to action at the bottom. The other runs from generatable, meaning a general-purpose model can reproduce the output for free, to bespoke, meaning it cannot, because the value sits in proprietary data, trust, or physical execution.
Companies selling nothing but access to answers a competitive model can generate sit in the most dangerous cell. That is where Chegg lived, alongside Course Hero and Stack Overflow.
Stack Overflow is the cleanest example of the pattern, because the moment a model answers the question directly, the site that monetized the question stops getting the click.
Zeile told me that in an interview, and the traffic numbers back him up. Even companies with elaborate products built on top of commoditized knowledge feel the same suspicion from investors. Duolingo’s stock has shed most of its value even as daily active users and paid subscribers kept climbing, a sign the market is pricing the durability of the business model rather than the strength of any single quarter.
Getty and Shutterstock’s Deal Dies in London
The two stock photo giants signed their merger agreement on January 6, 2025, betting combined scale would help them survive a market being hollowed out by AI image generators and cheaper licensing. The U.K. Competition and Markets Authority had other plans.
Regulators demanded Getty sell off Shutterstock’s editorial arm, the operation behind the Rex Features, Splash News, and Backgrid brands, before approving the tie-up. Getty declined. The companies let their extended deadline lapse on July 6, 2026, and terminated the agreement the next day. The U.S. Department of Justice had separately sent both companies a second request for information, adding scrutiny neither side wanted to sit through.
The stock photo market has been compressed toward a fraction of its former size, and the merger was supposed to be the industry’s shared answer to that squeeze. Instead, each company now faces the same AI pressure alone.
Art Zeile Has a Stake in This Story
Zeile is not a neutral analyst describing a trend from a distance. He runs DHI Group, the New York Stock Exchange company (DHX) behind Dice, a general tech job board, and ClearanceJobs, a marketplace for candidates who hold government security clearances.
“Nobody wants to pay for answers anymore, since the model hands those out for free, so what a company will pay for is a loop it owns, built on its own proprietary data, with a person accountable for the outcome,” Zeile told me. It is a sharp diagnosis. It is also, conveniently, a description of what DHI Group’s entire business is built to sell.
“Everyone is suddenly obsessed with forward-deployed engineers, and it is no accident, because that role sits exactly where the value is moving, inside the customer, close to the proprietary data, owning the result,” he said. Every company hunting for that kind of hire is a potential customer for Dice or ClearanceJobs.
Dice’s Own Numbers Tell the Same Story
DHI Group’s first quarter of 2026 shows the framework working on its own author. Total revenue came in at $29.7 million, down 8% year over year. The split between its two segments is where the pattern gets sharp.
| Segment | Q1 2026 Revenue | Year-Over-Year Change | Full-Year 2026 Guidance |
|---|---|---|---|
| Dice (general tech job board) | $15.7 million | Down 17% | $62 million to $64 million |
| ClearanceJobs (vetted, cleared candidates) | Included in total | Up 5% | Included in total guidance |
| DHI Group total | $29.7 million | Down 8% | $126 million to $128 million |
Dice, the segment that looks most like a generic marketplace for “who is hiring,” is shrinking fast. ClearanceJobs, the segment that requires a security clearance and sits closer to vetted, bespoke access, is growing. DHI Group’s own quarter is a small-scale replay of the map Zeile describes to the rest of the industry.
The stock still closed lower the day the results came out, at $2.48, down about 2%. Investors read the segment split the same way Zeile reads Chegg’s.
Is AI Killing All Freelance Work?
AI is replacing routine, well-scoped freelance tasks fast, but the destruction is not total. Firms that spend the most on AI overall are still adding headcount elsewhere, which looks more like a reallocation of work than a clean wipeout of jobs.
Ramp, the corporate card and spend management company, pulled transaction data across its client base and found labor marketplace spending sliding from 0.66% to 0.14% of total spend. More than half the companies that used freelancers in 2022 have stopped entirely. Among the heaviest former freelance spenders, every dollar shifted to AI tools replaced roughly $33 of freelance labor.
Jasper, Tome, and similar AI-native tools sit one square down from freelancers on the grid, closer to infrastructure, and they are not safe either. The moment a foundation model ships the same feature natively, sitting lower in the stack stops being protection.
The same Ramp data complicates a purely bleak reading. Its research found firms spending the most on AI are adding jobs even as they cut freelance budgets, evidence that money saved on gig labor is landing somewhere else in the org chart rather than disappearing outright.
Where Companies Are Racing to Hide
The bottom-right corner of the grid, bespoke work a model cannot reproduce, is getting crowded fast. Three moves show what racing there looks like in practice.
- Shutterstock now sells training data to the same AI models that shrank its stock library business, turning a former threat into a customer.
- Fiverr restructured around four pillars aimed at climbing from commodity gigs into vetted, higher-trust enterprise work.
- DHI Group is leaning into ClearanceJobs and forward-deployed hiring demand, the exact segment growing while its generic job board shrinks.
Dice’s own research shows the labor market catching up. AI skills appeared in 75% of tech job postings in June 2026, up sharply from a year earlier, with roles tied to agentic AI and AI infrastructure growing fastest of all.
“Hiring has come back, but the demand shifted toward AI-native developers who can put these tools to work inside a business, and honestly we still do not have enough people who can do that,” Zeile said. Yet DHI Group’s own report on tech professionals describing a defensive job market complicates that upbeat framing. Demand for specific AI-native skills is real. A broad, comfortable rebound for tech workers generally is not confirmed yet.
DHI Group’s own full-year guidance points to Dice revenue near $63 million, a job board selling access to postings and resumes, the same square on the grid Zeile calls most exposed. He is building the case for how other companies escape it. His own segment has not escaped yet.
Frequently Asked Questions
What Is a Forward-Deployed Engineer?
A forward-deployed engineer is a technical hire who works embedded inside a client’s own operation, building the specific data pipeline or workflow that company needs rather than shipping one generic product to everyone. The role became closely associated with Palantir before spreading across the wider software industry as companies chase the proprietary, hard-to-copy loops Zeile describes.
Is Chegg Going Out of Business?
Chegg is still operating and has not shut down, even after two rounds of layoffs and a sharp revenue decline. The company is now pushing into skills training and language learning, betting those markets hold ground its homework-help business could not.
Why Did United States Regulators Also Scrutinize the Getty-Shutterstock Deal?
Even as the U.K.’s Competition and Markets Authority forced the actual termination, the U.S. Department of Justice had sent Getty and Shutterstock a second request for information, a formal step that extends antitrust review and shows regulators on both sides of the Atlantic saw competitive risk in combining the two largest stock-image libraries.
Which Kinds of Businesses Are Safest From AI Disruption?
Under Zeile’s framework, the safest position is owning a proprietary loop a model cannot copy, one built on data the company alone holds, judgment a client trusts, relationships, or physical execution. Vetted marketplaces, embedded technical services, and businesses where a human stays accountable for the outcome all sit closer to that corner than any company that just hands over an answer.
