Mark Zuckerberg's Superintelligence Plan: What Meta Is Building and Why Advertisers Should Watch

Mark Zuckerberg once wrote a 6,500-word letter about the future of superintelligence and, according to one trade publication’s count, did not mention advertising a single time (PPC Land).

Which is funny, because advertising is how Meta pays for all of it. Meta’s ad revenue was $59.4 billion in the second quarter of 2026 alone (Variety).

So here is the curiosity gap: how does a plan for “personal superintelligence” end up inside your Meta Ads account? Connect three dots (the letter, the model and the ad tools) and the answer gets clear. Here is the plan, decoded.

How we wrote this: facts are linked to a source. Where we interpret, we say “our read.” Rumors are labeled. We could not open some primary pages directly while researching, so we rely on the reporting we link and flag where that matters.

The short version

Zuckerberg wants Meta to build an AI that knows you deeply and works for you, delivered mostly through smart glasses. Meta has reorganized around that goal, shipped its first model from the new unit, and is pushing AI hard into advertising. For advertisers, the practical effect is more automation inside Meta’s tools, and less control over the knobs you used to turn.

What is “personal superintelligence”?

It is Zuckerberg’s term for AI that understands each person’s goals and helps them reach them. In a July 2025 letter, he laid out the philosophy as “individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety” (CNBC).

The hardware part is explicit. Per coverage of the letter, he wrote that “personal devices like glasses that understand our context because they can see what we see, hear what we hear, and interact with us throughout the day will become our primary computing devices.”

He also made an argument about power: if superintelligence is concentrated in a few hands, outcomes are worse for everyone else. Critics note it is a convenient position for a company that wants to sell the glasses and the model access. That is a fair point, and we cannot see inside anyone’s motives. Our read: the vision is sincere and also very good for Meta’s business. Both can be true.

What has Meta actually shipped?

The headline product so far is Muse Spark, the first model from Meta Superintelligence Labs, which launched in April 2026. CNBC reported it as Meta’s first major model since the $14 billion deal that brought in Alexandr Wang to lead the lab (CNBC). TechCrunch described it as a “ground-up overhaul” of Meta’s AI (TechCrunch).

Here is the reality check, separating what exists from what is promised:

We did not verify Muse Spark’s exact benchmark standing or later version numbers from a primary source, so we are leaving them out.

What does the money tell us?

Meta is spending at a scale that only makes sense if it expects AI to pay back through ads. Per Variety, Meta’s second-quarter 2026 revenue was $60.8 billion, up 28%, and it narrowed its 2026 capital spending guidance to $130 billion to $145 billion (Variety).

Our read: when a company spends that much on compute, it needs a revenue engine already running. Meta’s is advertising. That is why AI keeps showing up in the ad tools first, long before “superintelligence” reaches consumers.

What is Meta doing to advertising with AI?

Meta wants advertisers to give it a goal and a budget, and let AI do the rest. Marketing Dive reported, citing earlier reporting from The Wall Street Journal, that Meta aimed to let brands provide a product image and budget and have AI create and target the ads (Marketing Dive). The expected timeline was the end of 2026.

What exists today, per industry coverage (MM+M):

• Advantage+ automates audiences, placements and budgets, and now sits in a merged campaign flow.
• Generative creative tools can turn product photos into video ads and apply other AI edits.
• Business AI agents that learn from a brand’s posts and website to recommend products inside Meta’s apps.

Treat the fully automated, URL-to-campaign version as in testing, not guaranteed. Some vendor blogs describe it as nearly here. We have not seen Meta confirm a general release date.

What could slow Meta down?

Three things: execution, trust and economics. None of them is fatal, but all of them matter if you are planning around Meta’s roadmap.

• Execution: Muse Spark is Meta’s first model from a new lab, competing with labs that have shipped for years. A strong debut does not guarantee a lasting lead.
• Trust: an AI that “knows you deeply” needs a lot of personal data. Meta’s track record on privacy will be scrutinized more, not less, as the assistant gets more personal.
• Economics: Variety also noted a $2.4 billion legal charge in the same quarter (Variety), a reminder that regulation and litigation are real costs, and that spending this much on infrastructure leaves little room for a weak quarter.

Our read: bet on the direction (more automation, more AI at the center), but do not bet your whole plan on any single date from a keynote or a news report.

Fear, gain and pain for advertisers

• The fear: a black box that spends your budget on creative you never approved, in a style that drifts from your brand, with reporting you can only check inside Meta.
• The gain: cheaper creative production and faster testing. If you have a great offer and clean data, automation can multiply it.
• The pain: your advantage moves. Targeting tricks are fading. What differentiates you now is your offer, your creative ideas and the quality of the signals you send back to Meta.

We covered the creative side in why Meta ads creative is now your targeting. Automation makes that argument stronger, not weaker.

Is your Meta account ready for automation? A checklist

• Conversion tracking is solid (server-side events where possible; see our Conversions API guide)
• You have a documented offer, brand rules and “never say” list the AI should follow
• You have a library of real product photos, UGC and proof, not only stock images
• You test hooks and angles on purpose, not by accident (our free Hook Bank helps)
• You keep a manual or controlled campaign to compare against automated ones
• You judge results on your own revenue and margin, not only Meta’s reported ROAS

What this means for your business

If you do not buy Meta ads, the signal is still useful. The biggest platforms are reorganizing around AI that acts for the user and automates the seller side. Both sides of the market are getting an assistant. The brands that win will be the ones the assistants can understand and trust.

If you do buy Meta ads, expect more of your campaign to be decided by the system. Our job, and yours, shifts toward feeding it better inputs and holding it accountable. That is the core of our Meta Ads service today.

What to do this week

1. Audit your inputs: list your top 10 product photos, proof points and offers. If it takes more than an hour, that is the gap.
2. Check your event quality in Events Manager and fix any broken or duplicate conversions.
3. Write a one-page brand rules doc for AI-generated creative: tone, claims you can and cannot make, and visuals to avoid.
4. Set up a holdout test: keep one manual campaign alongside an Advantage+ one and compare cost per real customer.
5. Download the free Hook Bank and brief three new creative angles to test next week.

Zuckerberg’s letter may never mention advertising. Meta’s business plan, meanwhile, clearly does.

---

WRITTEN BY

Junaid Mubashar

Founder of an AI-native growth agency. Junaid plans and runs full-funnel marketing (SEO and AI search, paid social, Google Ads) and builds the AI workflows and agents that automate it. Everything on this blog is tested with real campaigns before we recommend it.

About JunaidLinkedInYouTube

Free Swipe file · PDF

🪝 150 Scroll-Stopping Hooks

Fill-in-the-blank hook formulas for Reels, TikTok and Meta Ads — sorted by angle, each with a worked example.

Company website

Get it free →