When Alex Kantrowitz threw down his assessment of Meta’s AI strategy this week, he wasn’t just critiquing a product launch timeline or comparing feature sets. He was questioning whether one of the world’s most powerful tech companies understands the fundamental shift happening right beneath its feet. And his argument hinged on a single, uncomfortable number: 900 million weekly active users on ChatGPT versus what he called Meta’s “virtually nothing” in terms of intentional AI engagement.
I’ve watched this tension build for months now, sitting in on earnings calls and tracking how executives frame their AI progress. Meta announced nearly a billion monthly actives for Meta AI during its Q1 2025 earnings, a figure that sounds impressive until you start pulling it apart. The problem isn’t the number itself but what it represents. Meta AI lives inside Facebook, Instagram, WhatsApp, and Messenger, platforms where billions of people already spend their time. Encountering Meta AI isn’t a choice for most users; it’s more like bumping into a feature while you’re doing something else. ChatGPT users, by contrast, deliberately open an app or navigate to a website because they want to use AI. One is ambient; the other is a destination. That distinction matters enormously when you’re trying to figure out who’s actually winning the consumer AI race.
According to research from MIT Technology Review, the shift toward intentional AI usage signals a deeper behavioral change in how people interact with technology. When users actively seek out an AI tool rather than passively encounter it, they’re forming habits that could define the next decade of computing. Meta’s challenge isn’t just about building a better chatbot; it’s about creating a product compelling enough to pull people away from established routines. Right now, the data suggests they haven’t cracked that code.
Kantrowitz’s critique goes beyond user metrics and into what he calls the foundational model problem. Meta has invested staggering amounts of capital into AI infrastructure, with capex projections for 2026 alone ranging between $115 billion and $135 billion according to company guidance. Yet the company still hasn’t delivered a foundational model that competes directly with OpenAI’s offerings. Reports circulating in tech circles suggest Meta’s Avocado model rollout has been pushed to May or later, and prediction markets currently assign only a 10.5 percent probability to the Mango model releasing by the end of March. For a company spending more on AI than perhaps any corporation in history, these delays are more than logistical setbacks; they’re strategic vulnerabilities.
What strikes me most about this situation is how it reflects a broader tension in the tech industry right now. Meta’s advertising business remains extraordinarily strong, with Q4 2025 revenue hitting $59.89 billion, up nearly 24 percent year over year according to their latest earnings report. Earnings per share came in at $8.88, beating analyst estimates by over 8 percent. The machine that built Meta’s empire keeps humming along, generating cash at a pace most companies can only dream about. But operating margins compressed to 41 percent from 48 percent a year earlier, a sign that costs are accelerating faster than revenue. Investors are starting to ask whether all that AI spending will ever translate into competitive advantage or just become an expensive footnote in tech history.
The stock market’s reaction tells part of the story. META shares are down about 7 percent year to date, trading well below the 52-week high of $795.06. Analyst consensus sits around $862.25, a target that assumes Meta’s AI investments eventually pay off in meaningful ways. But that assumption is increasingly being questioned by people like Kantrowitz who cover this industry closely. As a CNBC contributor and founder of Big Technology, his perspective carries weight among investors trying to navigate the chaotic landscape of AI competition.
I spoke with several developers and AI researchers over the past few weeks, and a common theme emerged: intentionality matters more than reach when you’re building a new computing platform. One engineer at a major tech company told me that embedding AI features into existing products can boost engagement metrics, but it rarely creates the kind of user dependency that defines transformative technologies. People don’t think about the autocomplete in their email as AI; they think about ChatGPT when they want to solve a problem or create something new. That perception gap is Meta’s fundamental challenge.
Kantrowitz’s suggestion that Meta should look outside for solutions if internal development isn’t working represents a striking departure from the company’s historical playbook. Meta has always prided itself on building core technologies in-house, from the News Feed algorithm to its advertising infrastructure. Acknowledging that the company might need external help with AI would be a significant strategic pivot, one that raises questions about whether the current approach is fundamentally flawed or just taking longer than expected.
What makes this moment particularly fascinating is the contrast between financial performance and strategic positioning. Meta is printing money from its advertising business while simultaneously struggling to establish relevance in what many believe will be the next major platform shift. The company isn’t failing by traditional metrics; it’s facing an existential question about whether it can transition from being incredibly good at one thing to being competitive in something entirely different. History suggests that’s one of the hardest challenges in technology, and not every company succeeds.
The comparison to OpenAI also highlights different approaches to AI deployment. OpenAI built a standalone product that users had to discover and choose, creating genuine demand rather than manufacturing engagement through distribution. Meta’s strategy leverages its enormous existing user base to expose people to AI features, hoping familiarity will eventually turn into preference. Both approaches have merits, but only one so far has produced what Kantrowitz calls “intentional” users, and that matters when you’re trying to build the foundation for future computing platforms.
As 2026 unfolds, Meta faces a critical test. Can a company spending more on AI infrastructure than perhaps anyone in history actually produce a consumer product that people actively want to use? Or is all that capital going toward building what Kantrowitz memorably described as “a toll road with no cars on it”? The answer will determine not just Meta’s position in AI but whether its current valuation and investor confidence can hold up under the weight of enormous spending with uncertain returns. Right now, the market doesn’t have a clear answer, and that uncertainty is precisely what’s making this one of the most compelling technology stories to watch this year.
