Meta is widening its enterprise AI ambitions beyond the business-facing agents it introduced in June, according to comments from CEO Mark Zuckerberg on the company’s second-quarter earnings call. The new message is that Meta does not just want to sell AI assistants to businesses using its apps; it also sees room to offer APIs, computing capacity, and software that originated inside Meta itself.

That matters because it reframes Meta’s AI monetization story. For years, the company’s business has been dominated by advertising, with some additional subscription revenue. Zuckerberg’s latest remarks, as reported by TechCrunch, suggest Meta is now testing a more diversified enterprise AI strategy that could eventually put it in closer competition with cloud vendors, model providers, and workplace software companies — even as it still leans on its massive advertiser base as the first distribution channel.

From business messaging to a broader enterprise stack

The immediate starting point is still the product Meta launched in June: a business-oriented AI agent designed to help companies with customer service, support, and routine operations. On the earnings call, Zuckerberg said Meta first plans to serve the businesses already operating across its platforms, especially advertisers using messaging apps and other Meta surfaces to reach customers.

According to TechCrunch’s report, Zuckerberg described these tools as a natural extension of Meta’s existing commercial relationships. The logic is straightforward: millions of advertisers and hundreds of millions of small businesses already use Meta’s platforms, so adding AI-driven customer interaction tools could be less about opening a brand-new market than deepening usage inside the one Meta already controls.

He also tied the revenue model to business outcomes rather than only software access. TechCrunch reported that Zuckerberg said Meta expects to get paid when it delivers results for those businesses, positioning the offering as an extension of ad-tech and performance marketing rather than a pure seat-based enterprise software product.

But the more notable shift in the call was the emphasis on what comes after those agents. Zuckerberg said Meta sees a “large enterprise opportunity” that includes APIs, business agents, potentially selling compute directly, and other services for large customers. That wording suggests Meta is exploring a layered strategy: applications on top, infrastructure underneath, and internal tooling potentially turned outward.

Why Meta is talking about enterprise now

The timing reflects two pressures at once. First, the AI buildout underway across Big Tech is expensive, particularly around data centers and chips. Second, investors increasingly want to know how those investments turn into revenue beyond consumer engagement or ad targeting.

In that context, Meta’s enterprise comments look like an attempt to show that AI spending can support more than internal product improvements. If Meta can convert parts of its AI stack into commercial offerings, it gains additional ways to justify capex while diversifying away from dependence on advertising.

TechCrunch reported that Zuckerberg explicitly acknowledged enterprise selling is not a historic strength for Meta, calling it a “different muscle.” That admission is important. Meta has deep experience building consumer products and monetizing attention through ads. Selling into enterprise procurement, supporting large accounts, meeting security and compliance expectations, and competing in longer software buying cycles are all different disciplines.

Even so, Meta has two structural advantages. One is distribution through Facebook, Instagram, WhatsApp, and Messenger, where businesses already interact with customers. The other is internal scale: Meta builds coding, developer, and productivity systems for itself, and Zuckerberg suggested those tools could later be offered externally once they are sufficiently proven inside the company.

That is a familiar route in enterprise software. Internal tooling has often become a commercial product when vendors believe their own operational needs forced them to build something valuable. What is still unclear here is how productized those systems are, what customer segments Meta would target first, and whether buyers would treat Meta as a serious software supplier beyond marketing and messaging use cases.

Compute may be the most consequential piece

Of all the opportunities Zuckerberg mentioned, selling compute may carry the biggest strategic implications. TechCrunch reported that Meta said it currently has chances to sell compute at a “significant premium” to what it paid. In a market defined by GPU scarcity, that is a notable signal.

But Meta also made clear it is not eager to maximize short-term infrastructure revenue at the expense of its own roadmap. Zuckerberg reportedly said it would be “foolish” to sell all available compute for immediate profit, describing Meta’s infrastructure approach as a portfolio balancing near-term and long-term needs.

That balance matters because Meta is simultaneously pursuing large ambitions in AI systems for both enterprises and consumers. On the same call, according to TechCrunch, Zuckerberg connected hardware decisions to Meta’s push toward what he called “personal superintelligence,” arguing the company will need hardware that supports seamless interaction with those systems.

For the market, this puts Meta in an unusual position. It is not a conventional cloud provider, but it has enough infrastructure spending and AI demand to consider acting like one in selective cases. Whether that becomes a durable business line or just opportunistic capacity sales remains uncertain.

The enterprise pitch sits inside a larger AI product strategy

Meta’s enterprise comments did not come in isolation. They sit alongside a much broader bet on agentic software, consumer AI, and faster app development enabled by large language models.

TechCrunch reported that Meta sees agentic AI not only as a business tool but also as a consumer product category, including “personal AI agents” and AI smart glasses. At the same time, Zuckerberg said large language models are helping Meta ship more software across its consumer ecosystem, including recent experiments for Marketplace sellers, Facebook Groups, and vibe-coded games.

That is relevant to enterprise buyers because it shows Meta is not building a stand-alone enterprise AI unit with a single focus. Instead, it appears to be creating shared AI capabilities that can feed multiple businesses: ad systems, business messaging, internal developer tools, new apps, and hardware experiences.

For builders, that can be attractive if it means rapid iteration and a large platform surface. It can also raise questions about priority. Enterprise customers usually want predictable roadmaps, support commitments, and stable interfaces. Meta’s culture has traditionally favored consumer-scale experimentation. Turning shared internal AI capabilities into dependable enterprise products is possible, but it is a different operational discipline from launching features across social apps.

Evidence, claims, and what remains unproven

The core facts in this story come from Zuckerberg’s comments on Meta’s second-quarter earnings call, as relayed by TechCrunch. The strongest claims about opportunity size, product direction, and revenue potential are therefore company statements, not independently verified market outcomes.

What is confirmed from the reporting is limited but meaningful: Meta launched a business AI agent in June; Zuckerberg says Meta now sees a larger enterprise opportunity spanning APIs, business agents, compute, and other services; and he says Meta may eventually offer internal coding, developer, and productivity tools to outside customers.

Other important parts are still claims or intentions. There are no disclosed customer counts for the new business agent in the source material. There are no reported pricing details for APIs or compute, no timeline for externalizing Meta’s internal tools, and no evidence yet that large enterprises beyond Meta’s existing advertiser base are already adopting these offerings at scale.

The compute economics also require caution. Meta’s statement that it can sell compute at a premium is company-reported commentary from the earnings call, not a disclosed contract pipeline. Likewise, the broader vision around “AI agents,” “APIs,” and “enterprise AI” describes a direction of travel more than a complete product catalog.

What this means for builders and enterprise buyers

For product teams and founders, Meta’s move is a reminder that the battle for enterprise AI is not just between model labs and cloud providers. A platform company with large distribution in messaging and advertising can enter from the workflow layer first, then expand downward into infrastructure and sideways into internal tooling.

For small businesses already using Meta channels, the immediate opportunity is practical: AI agents tied to customer conversations across Meta properties may be easier to adopt than a full enterprise software platform. If those tools are priced around performance, they could feel closer to ad spend than to traditional software procurement.

For larger enterprises, the picture is less mature. Interest in Meta as a vendor may depend on whether the company can prove reliability, governance, integration depth, and support quality for business-critical use cases. Buyers will also want to know how Meta separates advertising relationships from enterprise product relationships, especially around data handling and account management.

For developers, any future Meta APIs or externally offered internal tools would be worth watching because they could expose capabilities built at enormous internal scale. But until Meta publishes concrete products, availability, and documentation, the current announcement is best read as a strategic signal rather than a finished enterprise platform launch.

What to watch next

The clearest follow-up signal will be product detail. Watch for Meta to specify whether its enterprise push appears first through WhatsApp, Messenger, Facebook, or a separate admin layer for businesses.

A second signal is packaging. If Meta starts naming paid APIs, publishing usage-based pricing, or describing service levels, that would mark a shift from earnings-call strategy to deployable enterprise offer.

Third, watch whether Meta says more about externalizing its internal developer stack. If coding, productivity, or developer tools become standalone products, Meta would be competing more directly with workplace software and coding assistant vendors rather than only with customer support automation platforms.

Finally, infrastructure disclosures will matter. Any clearer statement about compute sales, customer types, or how Meta allocates capacity between internal AI systems and outside demand would help determine whether compute is a side business or a strategic revenue stream.

Creati.ai perspective

Meta’s pitch is notable less for the business agent it already launched than for the monetization map now forming around it. The company appears to be using business messaging as the near-term wedge, while keeping open the option to sell infrastructure and internal software later. That is a sensible sequence because it starts where Meta already has distribution and measurable business outcomes.

The harder part will be execution. Enterprise AI rewards reach, but it also rewards trust, support, and boring operational consistency. Meta clearly has the technical scale to matter in enterprise AI, and its comments on compute and internal tools show ambition beyond chatbot features. The unanswered question is whether buyers will view Meta as a long-term enterprise platform partner or mainly as a powerful channel for customer-facing AI interactions. That distinction will shape how far this opportunity extends beyond agents.