Zuckerbergs WSJ Op-Ed, Muse Spark, and the Fight Over Who Controls AI

Then came the line everyone actually noticed. "It is surprising that the discourse from many of those who are developing artificial intelligence is so filled with doom. I don't understand why anyone who believes that AI will eliminate most jobs and much of humanity's relevance would rush to build that future," he wrote. His broader argument: the history of democracy and economics has proved that centralized power stifles human potential.

Nobody named names. Nobody had to. Within hours the piece was being read as aimed squarely at Anthropic's Dario Amodei, the industry's most prominent doom-adjacent voice and the executive who has spent 2026 publicly warning that AI could eliminate half of entry-level white collar jobs. One widely shared reaction on X described it as Zuckerberg landing "an elbow to Dario's face." Whether or not that was the intent, the framing writes itself: Meta as the democratizer, Anthropic as the gatekeeper hoarding a dangerous technology "for your own good."

This isn't a new pose for Zuckerberg. A year earlier he wrote almost the identical line, "Meta's vision is to bring personal superintelligence to everyone. We believe in putting this power in people's hands to direct it towards what they value in their own lives," and even then it was read as shade thrown at rivals who "believe superintelligence should be directed centrally towards automating all valuable work." What's different this time is the product underneath the rhetoric actually exists now, and the timing is not an accident. $META reports Q2 2026 earnings on July 29, with Wall Street expecting EPS of roughly $7.18 on revenue of $60.22 billion. Publishing a manifesto about why your AI strategy is the righteous one, the day before you have to defend $145 billion of capex to analysts, is not a coincidence, it's a page from the same playbook Zuckerberg ran with his original superintelligence letter, which also landed the same day as Meta's Q2 earnings call the year before.

Here's where the "hints about superintelligence through Facebook, WhatsApp and Instagram" part gets concrete. In April, Meta shipped Muse Spark, created by Meta Superintelligence Labs, the team of researchers led by Chief AI Officer Alexandr Wang, and a pivot from the company's prior open source strategy, since Muse Spark is a closed model whose design and code won't be made public. It currently powers the Meta AI app and website, with plans to roll out across WhatsApp, Instagram, Facebook and AI glasses. In July, Meta pushed it further with Muse Spark 1.1, which Wang said stands out for strong tool use, a million token context window, parallel subagents and advanced computer use capabilities, and for the first time Meta is actually charging for it: per token fees at roughly 25% of rival pricing, marking Meta's first real push to monetize AI models.

The pitch for why this beats a smarter, closed research lab model was spelled out plainly at launch. "The future of Meta AI is rooted in the relationships and context already at the center of your life," the company said. "We are building toward personal superintelligence, an AI that does not just answer your questions but truly understands your world because it is built on it." That's the actual strategy hiding inside the WSJ op-ed's talk of democratization: Meta doesn't need to win the model quality race if it wins the distribution race, because it already has billions of people inside Instagram DMs and WhatsApp threads that a closed lab like Anthropic has to acquire from zero.

Money is the other half of why this op-ed exists. $META closed Monday at $593.87, well off its 52 week high of $796.25, and the stock is down about 10% so far in 2026, weighed down by child safety verdicts, a $375 million fine, and AI spending anxiety. Its 52 week low of $520.26 marks how deep that stretch went. The launch of Muse Spark from its Superintelligence Labs has revived some optimism around the company's AI roadmap. The capex number is the real tell of how seriously Zuckerberg is playing this: Meta raised 2026 capex guidance to $125 to $145 billion, up from a prior $115 to $135 billion range, citing higher component pricing and data center costs, with contractual commitments stepping up $107 billion this quarter for multiyear cloud and infrastructure deals. For scale, the company's last full reported fiscal year ran $69.7 billion of capex against $46.1 billion of free cash flow. Analysts are mostly still buying the story: of 71 ratings, 63 are buy or strong buy and not one is a sell, and the average price target sits near $826 across 58 targets.

The Dario Amodei subplot has its own separate, uglier front, and it predates this week's op-ed by months. Zuckerberg spent last summer poaching Anthropic, OpenAI and Google researchers with eye-watering packages. Amodei has been publicly needling him about it since, and mostly claiming victory. Amodei said Anthropic is generally faring better than the competition amid Zuckerberg's efforts to poach top talent, with reported pay packages as high as $100 million, adding "you can see publicly the list of people who went to the Meta Superintelligence Lab. Even if you normalize for our size, many turned them down." Somewhere between that talent war and this week's op-ed, the tone shifted from a business rivalry into something closer to open contempt. Amodei has accused Meta of luring away his engineers with billion dollar signing packages and called separate remarks about Anthropic from Nvidia "an outrageous lie."

Worth separating two things that get blended in the discourse: Zuckerberg's op-ed is a business and philosophy argument. The much nastier fight Amodei is currently in is with Washington, not Menlo Park, and it's a different kind of ugly. Emil Michael, the undersecretary of defense for research and engineering, posted on X that "it's a shame that Dario Amodei is a liar and has a God-complex," and Defense Secretary Pete Hegseth called Amodei "an ideological lunatic" during an April congressional hearing, after the Pentagon labeled Anthropic a supply chain risk. Zuckerberg didn't say any of that. But his op-ed lands in the same news cycle where the industry's most prominent safety voice is simultaneously being attacked from the White House and getting quietly out-recruited by the guy writing "centralized power stifles human potential" op-eds. The optics do the work the words don't.

I run a software company that has tried to actually live the "distribute the power" side of this argument, a small team leaning hard on AI leverage instead of headcount. The honest update after a year of doing that for real: capability existing somewhere doesn't automatically become distributed power. Someone still has to wire it into an actual workflow, and most of the value shows up in narrow, unglamorous places, not in a 616 word manifesto about personal empowerment. Zuckerberg's bet that owning WhatsApp and Instagram beats owning the smartest model is genuinely testable, and it's testable in the numbers, not the prose.

Two things worth checking yourself rather than taking anyone's word for it. You can see exactly how aggressive Meta's spending really is against its own cash generation in Meta Platforms capital expenditure versus free cash flow trend, and you can check whether Wall Street's optimism about Muse Spark is actually showing up in Meta Platforms analyst price targets and earnings revisions.

Zuckerberg will get his answer on whether the market buys "distributed superintelligence" as an investment thesis in about 24 hours, when Meta reports Wednesday afternoon. The op-ed was never really written for Dario Amodei. It was written for the people about to grade a $145 billion bet.