Who Is AI Sending Traffic To? The Hidden Winners Beyond Big Tech

March 11, 1:22 pm

The AI buzz has reached a fever pitch. Every company in Silicon Valley claims to be an AI company. Every earnings call drops the term dozens of times. But behind the hype sits a simple, measurable question: which companies are AI tools actually sending people to?

At AltIndex, we track web traffic across thousands of publicly traded companies. For this analysis, we focused specifically on referrals from large language models — including ChatGPT, Claude, Perplexity, and similar AI tools — during February 2026. We excluded the obvious names. Google, Meta, Amazon, Microsoft, and Apple are all deeply embedded in the AI ecosystem through partnerships with frontier labs or their own models. Their AI-driven traffic is expected, well-covered, and already priced into their stocks.

What we wanted to find were the companies investors might be overlooking. The ones where AI referrals are quietly becoming a meaningful share of total web traffic. The results turned up some genuine surprises.

---

Setting the Scene

LLM referral traffic is still a small slice of total web visits for most companies, typically 0.1% to 0.7% of overall volume. But the absolute numbers are not trivial, and the share figures are rising fast enough to matter. Here is the full picture from February 2026 across the companies we analyzed:

Reddit (RDDT): The Internet's Most Cited Human Voice

Highest raw volume  ·  6,562,000 LLM visits in February

No company in our dataset comes close to Reddit on raw LLM referral volume. ChatGPT, Claude, and Perplexity collectively sent more than 6.5 million visits to Reddit in February alone. That is a reflection of something structural: Reddit is one of the primary sources LLMs cite and link to when answering questions about products, experiences, opinions, and recommendations. When people ask an AI model what a product is actually like to use, they tend to get a link to a Reddit thread.

The logic is straightforward. Reddit's content is entirely human-generated, community-verified, and hard to replicate. For AI models competing on answer quality, Reddit threads remain among the most credible sources on the web. At its peak, Reddit was being cited in more than 14% of ChatGPT responses, according to data from Promptwatch. That share has since fallen dramatically — sending the stock down 11.9% in a single session when the drop was first reported in October 2025. By February, citations had stabilized, but the episode highlighted just how directly Reddit's AI relevance can move its stock price.

That story plays into what makes Reddit one of the most interesting setups in our dataset. The fundamental business has never been stronger. Full-year 2025 revenue hit $2.2 billion, up 69% year over year. Reddit recorded its first full year of GAAP profitability, with net income of $530 million. Daily active users grew 19% to 121 million. The company even approved a $1 billion share buyback. By any conventional measure, this is a company hitting its stride.

And yet the stock is trading at $137, roughly 49% below its all-time high of $270. The bear case is that AI will eventually summarize Reddit's content without linking to it, slowly eroding the referral traffic that supports its advertising business. The bull case is that human-generated, moderated conversation is the one thing LLMs cannot fabricate — and that Reddit is actively monetizing its data advantage through licensing deals with OpenAI and Google. Analysts covering the stock hold a median price target around $205, implying about 50% upside from here.

---

Coursera (COUR): The Highest AI Share

Highest LLM share  ·  0.69% of all traffic from AI

If Reddit wins on raw volume, Coursera wins on something more telling: share. Nearly 0.69% of all Coursera's February traffic came from LLM referrals — the highest figure in our entire dataset, more than double Adobe or Pinterest, and four times what Spotify sees.

The reason is intuitive. When someone asks an AI model how to learn Python, what the best data science certification is, or how to get a Google Cloud credential, Coursera is one of the most frequently recommended destinations. Its courses from top universities and major tech companies sit at the intersection of exactly what people ask LLMs about: skills, certifications, and career reinvention. That kind of organic top-of-funnel does not show up in marketing spend — it is earned.

The strategic picture got considerably more interesting in December 2025, when Coursera announced an all-stock merger with Udemy. The combined company would have pro forma annual revenue of more than $1.5 billion, anticipated cost synergies of $115 million within two years, and a market value around $2.5 billion. Both companies cited AI prominently in the rationale: as artificial intelligence reshapes job requirements faster than traditional education can respond, the demand for reskilling is growing faster than either platform can serve alone. The deal is expected to close in the second half of 2026, pending regulatory and shareholder approvals.

---

Shutterstock (SSTK): The Merger Arbitrage Play With an AI Twist

2nd highest LLM share  ·  0.50%  ·  CMA decision due April 19

Shutterstock sits in a genuinely unusual position. It is one of the companies most frequently cited by LLMs when users ask about stock photography, creative assets, or image licensing. Half a percent of its total traffic came from AI referrals in February, second in our share ranking. And yet the same AI revolution driving that traffic is the biggest structural threat to its business model.

Shutterstock's response has been to lean into the disruption. The company has forged data licensing partnerships with OpenAI, NVIDIA, Meta, and Runway, selling access to its library of over 400 million images for AI training. One analysis projected AI image licensing could account for roughly a fifth of Shutterstock's total revenue by 2027. That is a meaningful hedge — but it is also self-cannibalizing. Every image trained on becomes ammunition for generating synthetic substitutes that eventually compete with Shutterstock's core product.

The more immediate question is the proposed Getty Images merger. The DOJ cleared the deal unconditionally in February 2026, but the UK's Competition and Markets Authority has raised concerns about competition in the UK editorial image market. The CMA issued its interim report on February 19, provisionally finding no global stock content issues but flagging a potential substantial lessening of competition in UK editorial. A final decision is due by April 19, 2026.

---

Booking.com (BKNG): Down 32%, Stock Split Coming, AI Fears Fading

904,000 LLM visits  ·  25-for-1 split effective April 6, 2026

The conventional worry about Booking.com is that AI travel agents will eventually disintermediate online travel agencies entirely. Why click through to Booking if an AI can just handle the booking for you? It is a reasonable concern. Alphabet launched an update to its AI Search Travel Mode in late 2025 that lets agents book trips inside the Google ecosystem, and the threat has helped push BKNG down 32% from its 52-week high.

But a few things happened recently that shifted the narrative. OpenAI scaled back its in-chat checkout plans for ChatGPT, easing a significant near-term disintermediation fear. And Booking's own Q4 2025 earnings report was genuinely strong: revenue rose 16% to $6.3 billion, beating estimates, while free cash flow surged 120% to $1.42 billion as operating margins expanded to 34.6%. Booking also disclosed that its agentic AI tools reduced average cost per booking by roughly 10% in 2025 — meaning AI is currently a tailwind for Booking's margins, not a headwind.

The company then announced its first-ever 25-for-1 stock split, effective April 6, 2026. This will bring the share price from around $4,300 to roughly $170, meaningfully broadening retail accessibility. Research from Bank of America has shown that companies announcing forward splits have historically outperformed the S&P 500 in the 12 months following the announcement.

---

What the Data Tells Us

LLM referral traffic is still a small slice of total web visits. But it is directionally meaningful, and it tends to correlate with something harder to measure: how frequently an AI model considers a brand credible, authoritative, and worth recommending. In a world where organic search is being disrupted by AI-generated summaries that answer questions without sending users anywhere, earning a place in AI recommendations is a new form of brand equity.

The companies in our dataset fall into roughly three camps. Some, like Reddit, hold a structural content advantage that LLMs genuinely depend on — but face the risk that AI eventually summarizes rather than links. Others, like Coursera, benefit because they sit at the end of questions AI is being asked constantly — and that traffic signal is only likely to grow as AI reshapes the workforce. And some, like Shutterstock, face a paradox where AI is simultaneously their best new customer and their most dangerous competitor.

Then there is Booking — a company where AI fears have created a material valuation discount in a business that is, for now, using AI to become more efficient rather than being disrupted by it.

The traffic data does not predict stock prices. But it does surface questions worth asking. And right now, the most interesting question our data raises is which of these companies will still be capturing AI referrals — and profiting from them — many years from now.

Disclaimer: The information provided is for educational and informational purposes only and should not be construed as financial or investment advice. All investments involve risk, and you should conduct your own research or consult a qualified professional before making any investment decisions.

Get More Insights

Sign up and get access to a personalized dashboard, deeper insights, AI stock picks, stock alerts, weekly newsletter and much more.