According to CNET, during Meta’s Q4 2025 earnings call, CEO Mark Zuckerberg outlined an aggressive plan to build a “personal superintelligence” AI. The company expects capital expenditures to skyrocket from $72 billion last year to between $115 and $135 billion, primarily to fund its AI labs. Zuckerberg argued Meta’s key advantage is the “unique context” it has from the troves of personal data collected on Facebook, Instagram, and WhatsApp for decades. This data, covering user history, interests, and relationships, is what he believes will allow Meta AI to provide a “uniquely personal experience.” The plan involves merging large language models with the social media recommendation algorithms that build our feeds.
The Data Advantage Play
Here’s the thing: Zuckerberg isn’t wrong about the data. While Google knows what you search for and Microsoft knows how you work, Meta’s platforms are a non-stop feed of your personal life—your family, your friends, your political rants, your vacation photos, your hidden interests. That’s a profoundly intimate dataset for training an AI meant to understand *you*. The company’s stated goal is to use this to have AI “understand people’s unique personal goals” and tailor everything, including your feed, to meet them. It’s the logical, if somewhat chilling, endpoint of a business built on surveillance-based advertising. They’ve spent twenty years mapping the human social graph. Now they want to animate it with AI.
A Rocky Road And Fierce Competition
But Meta’s path hasn’t been smooth. Their 2025 AI development was a mess of high-profile hires and internal chaos. They poached top researchers from OpenAI and Apple, but reports of internal strife and conflicting strategy between new hires and the existing FAIR lab stalled major releases. It got so bad they laid off hundreds from AI units and saw pioneer Yann LeCun leave his chief scientist role. So, while Meta was tangled in internal politics, competitors were shipping. Google’s Gemini 3 showed off crazy reasoning, OpenAI pushed out GPT-5.2, and Anthropic’s Claude became a coder’s best friend. Google even started rolling out personalized intelligence in Search. Meta’s playing catch-up in the raw model race, which is why they’re leaning so hard on the data angle as their differentiator.
privacy-reckoning-waiting-to-happen”>The Privacy Reckoning Waiting To Happen
And this is where the big tension lies. Zuckerberg’s vision sounds powerful, maybe even useful. Who wouldn’t want an AI that truly gets them? But the foundation is that same data-hungry, “opt-out-is-impossible” model that’s sparked backlash for years. Remember the uproar over AI in WhatsApp and using AI chats for ad personalization? That was just a preview. The company is basically saying, “We know everything about you, and we’re going to use it to build your personal super-brain. Oh, and you can’t turn the training off.” It’s a massive bet that users will trade even more privacy for perceived convenience and personalization. Given Meta’s recent reorgs around “superintelligence”, they’re clearly doubling down. The question is, will the public go along for the ride this time, or is this the step that finally triggers a real regulatory or user revolt?
