Internet stocks spent 2025 watching everyone else have a better year. While the broader market chugged along nicely, internet companies struggled with execution problems, mounting competition from generative AI, and profit margins that bounced around like a pinball. Now heading into 2026, the question is simple but high stakes: can these companies actually turn AI from a buzzword into money?

JP Morgan analyst Bryan M. Smilek points out that internet stocks underperformed both the S&P 500 and the Russell 2000 last year. The culprits? Operational stumbles, generative AI threatening established business models, and margins that refused to follow a predictable path. Even with macro conditions staying relatively stable, tariffs and changes to de minimis trade rules kept consumer sentiment in check.

And consumer sentiment isn't exactly thriving. Although it ticked up about 5% month over month through early December, it's still down roughly 28% year over year and hovering near record lows. The recent government shutdown and tariff actions didn't help matters.

Show Me The Money, AI Edition

Here's where 2026 gets interesting. According to Smilek, a stable consumer environment by itself won't be enough to keep valuations afloat this year. Companies need to execute better and prove they can actually monetize AI while turning a profit doing it. No more hand-waving about potential.

In education technology, Smilek expects deeper integration with AI platforms to help diversify where traffic comes from, plus separate scaling of AI-enhanced learning products. The winners here will be companies that have enterprise tech integrations, personalized learning support, massive datasets, proprietary content libraries, and—most importantly—proven outcomes. Having AI features is table stakes. Having AI features that demonstrably work is the actual game.

Then there's agentic commerce, which Smilek expects to accelerate. This is already happening in real, measurable ways. AI-driven traffic currently makes up about 1% of Etsy, Inc. (ETSY) total traffic, while industrywide AI referrals exploded roughly 1,200% year over year in October. The conversion rate tells an even more compelling story: AI-driven traffic converted about 16% better than non-AI traffic. When AI actually sends you better customers who buy more stuff, that's not hype—that's a business model.

Who's Positioned To Win

Smilek has specific views on individual stocks, and they paint an interesting picture of who might thrive versus merely survive.

For Duolingo, Inc. (DUOL), investor sentiment is decidedly negative right now. But Smilek thinks early 2026 bookings and adjusted EBITDA guidance could surprise to the upside. The bar is clear: Duolingo needs to deliver more than 20% bookings growth, stabilize daily active users, advance product initiatives, and expand margins in 2026. It's a tall order, but Smilek maintains an Overweight rating, suggesting he thinks they can pull it off.

On Etsy (ETSY), sentiment is mixed. Smilek notes that if gross merchandise sales growth holds up, it could offset the margin pressure from investing in Depop. That could support stronger share performance in 2026, though the path isn't exactly straightforward.

The upgrade story belongs to Coursera, Inc. (COUR), which Smilek moved to Overweight. His argument is that consensus estimates are underestimating improving execution across product development, go-to-market strategy, and profitability. Plus, the pending merger with Udemy, Inc. (UDMY) should create a more competitive AI-skills platform and accelerate AI product development. When two education platforms combine forces in the middle of an AI skills boom, the timing could hardly be better.

For the full picture, Smilek's 2026 ratings shake out like this: Duolingo and Coursera are Overweight. Etsy and Udemy get Neutral ratings. Meanwhile, Nerdy Inc. (NRDY) and Chegg, Inc. (CHGG) are rated Underweight, suggesting they face steeper challenges ahead.

The bottom line is that 2026 is when internet companies need to stop talking about AI and start showing profits from it. The technology is real, the early data looks promising, and the market is clearly ready to reward companies that can execute. But patience for potential without performance is running thin. This year will separate the companies that can actually build AI-driven businesses from those just riding the narrative.