Speaking during the company’s Wednesday (April 29) earnings call, CEO Mark Zuckerberg said these tools were handling 10 million conversations a week by late March, compared to 1 million at the start of the year.

And while Meta has not begun monetizing its business AI tools yet as it offers them gratis to small businesses as it attempts to scale, the founder hinted that could change.

“Business AIs today are currently free for most businesses on our messaging apps, but as we make more progress, we expect that we will also work towards establishing a longer-term monetization model,” Zuckerberg told analysts.

Muse Spark, Meta’s first model from the company’s in-house artificial intelligence (AI) lab, debuted during the first quarter and now powers Meta AI across all apps and the standalone Meta AI app. Sessions per user grew double digits following the introduction, with more advanced models now in training.

Meta also rolled out Meta Ads AI Connectors in open beta, allowing advertisers to connect their ad accounts to external AI agents for campaign management and optimization.

Susan Li, Meta’s chief financial officer, also noted that use of the company’s creative tools was scaling as well, with upward of 8 million advertisers using at least one of these offerings. Adoption has been especially strong among small and medium-sized businesses.

“These tools are benefiting performance as well, with advertisers using our video generation feature seeing more than 3% higher conversion rates in tests,” Li said.

Meanwhile, Zuckerberg covered Meta’s long-term vision for AI during the call, framing personal superintelligence as the company’s chief product goal.

“My view is that human progress has always been driven by people pursuing their individual aspirations,” he said.

The company also upped its full-year guidance for capital expenditures, going from its earlier range of $115 billion to $135 billion to a new projection of $125 billion to $145 billion. Meta cited rising cost of memory and additional data center capacity, and acknowledged that it has consistently underestimated its compute needs.

As PYMNTS wrote in March, this level of spending puts Meta in the company of some of the biggest investors in AI infrastructure, including Amazon, Google and Microsoft.

Meta is also reportedly slashing jobs in a move that is widely viewed as an effort to offset the investments it has been making in AI infrastructure. Last week saw a report that the company would cut 8,000 jobs, and leave another 6,000 positions unfilled.

Meta is now leaning on larger, more advanced artificial intelligence systems to decide what users see and when. These systems ingest longer user histories and interpret richer signals across video, text and interactions. They make predictions in real time. In place of narrower ranking models, Meta is deploying LLM-scale architectures that process longer interaction histories and more detailed content signals.

The results are showing up across both sides of the business. Engagement is higher. Ad conversion rates are higher, according to first quarter earnings.

From Better Recommendations to Better Ads

Meta’s recommendation systems now train on longer records of what each user watches, clicks and skips. The result: Same-day posts represent more than 30% of recommended reels on Facebook and Instagram, more than double the share from a year ago. Total video time on Facebook grew more than 8% globally in Q1, the largest quarterly gain in four years from the first quarter earnings call.

The same logic applies to ads. Meta built a system that identifies which ad requests are most likely to convert, then routes those to more powerful models at the moment of serving. In Q1, that approach drove a 1.6% increase in conversion rates for off-site campaigns. A separate improvement to landing page view ads drove more than a 6% increase in conversion rates.

“We’re going to be able to develop a first principles understanding of what you care about and what each piece of content in our system is about,” CEO Mark Zuckerberg said on the earnings calls. “So that way we can show you more useful things for what you’re trying to accomplish.”

More than 8 million advertisers now use at least one of Meta’s AI creative tools. Those using the video generation tool saw more than 3% higher conversion rates in tests. The Meta AI Business Assistant, rolled out to all eligible advertisers, is resolving common account issues at a 20% higher rate than before.

Agents at Scale

Meta’s business AI, available on WhatsApp and Messenger, handled 10 million conversations per week in Q1, up from 1 million at the start of the year. The company is expanding to more countries in Q2.

Muse Spark, Meta’s first model from its in-house AI lab, launched in Q1. It now powers Meta AI across all apps and the standalone Meta AI app. Sessions per user grew double digits following the rollout. More advanced models are in training.

Meta also launched Meta Ads AI Connectors in open beta, letting advertisers link their ad accounts to external AI agents for campaign management and optimization.

Where the Business Is Shifting

Ad impressions grew 19% year over year in Q1. The average price per ad rose 12%, driven by better ad performance, improved macro conditions, and currency tailwinds. Family of apps ad revenue reached $55 billion, up 33%.

The value optimization suite, which steers ad spend toward higher-value conversions rather than volume, now runs at an annual revenue run rate above $20 billion, more than doubling year over year. Partnerships ads, which lets creators tag and earn commissions on products in their posts, reached a $10 billion revenue run rate in Q1, also more than doubling year over year.

Reality Labs revenue was $402 million, down 2% year over year. Quest headset sales fell; AI glasses growth partially offset the decline. Daily users of AI glasses tripled year over year. More than 500 million users on each of Facebook and Instagram watch AI-translated videos weekly.

Family daily active people reached 3.56 billion in March, down slightly from December due to internet outages in Iran and a WhatsApp block in Russia. Growth would have been positive quarter over quarter without those disruptions.

What Else Stood Out

• Zuckerberg outlined Meta’s long-term AI vision on the call, framing personal superintelligence as the company’s primary product goal. “My view is that human progress has always been driven by people pursuing their individual aspirations,” he said.
• Meta raised full-year capex guidance to $125 billion to $145 billion, up from $115 billion to $135 billion, citing higher memory pricing and additional data center capacity. The company said it has consistently underestimated its compute needs.
• WhatsApp paid messaging and subscriptions drove family of apps other revenue to $885 million, up 74% year over year. Business AI expanded to small and medium businesses across Latin America and Indonesia during the quarter.
• Ray-Ban Meta released a new optics line designed for all-day wear in Q1, with sales shifting toward the latest generation as buyers respond to improved battery life and video resolution.
• Meta’s contractual commitments stepped up by $107 billion in Q1, driven by multiyear cloud deals and infrastructure purchase agreements intended to secure capacity through 2027.

Topline Results and Outlook

Q1 2026 total revenue was $56.3 billion, up 33% year over year, or 29% on a constant currency basis. Family of apps revenue was $55.9 billion, up 33%. Operating income was $22.9 billion, a 41% operating margin. Net income was $26.8 billion, or $10.44 per share, including an $8.3 billion tax benefit. Capital expenditures were $19.8 billion. Free cash flow was $12.4 billion. Meta held $81.2 billion in cash and marketable securities at quarter-end.

China’s National Development and Reform Commission announced Monday (April 27) that it would “prohibit foreign investment in the Manus project,” and required the companies to “withdraw the acquisition transaction.”

PYMNTS has contacted meta for comment but has not yet gotten a reply.

The commission’s decision comes months after a report from CNBC that Chinese regulators were investigating the deal, with a focus on whether the acquisition is in keeping with the country’s export control laws.

Meta announced plans to acquire Singapore-based Manus late last year as part of a larger effort to bolster its AI offerings. 

“Manus is already serving the daily needs of millions of users and businesses worldwide,” Meta said in its announcement. “It launched its first general AI agent earlier this year and has already served more than 147 trillion tokens and created more than 80 million virtual computers. We plan to scale this service to many more businesses.”

The deal marked one of the most high-profile examples of an American tech giant buying an AI startup with roots in Asia’s AI and startup spaces.

Manus gained the support of the Chinese government in March of 2025 after it introduced an AI agent that could produce detailed research reports and create custom websites, working with AI models from companies like Anthropic and China’s Alibaba.

Acquiring Manus would give Meta a “scaled, revenue-generating AI product with direct consumer payments,” PYMNTS wrote in a separate Dec. 30 report. 

Although the tech giant has invested in AI infrastructure and promoted open-source models with its Llama family of models, monetization has been indirect, linked largely to advertising and engagement across social media platforms like Facebook and Instagram.

“By acquiring Manus, Meta gains technology and distribution, along with immediate exposure to subscription revenue and insight into consumer willingness to pay for AI-powered assistance,” the report added. “The transaction also shortens the timeline for rolling out premium AI offerings without having to build a paid user base from scratch.”

In related news, PYMNTS wrote earlier this month about the way Meta’s status as a social media platform gives it an edge over other players in the AI space, thanks to years of accumulated public user data.

“No other AI company holds that position,” that report said. “OpenAI knows what users have asked previously. Google knows what they search. Meta knows what they buy, who they follow and what they scroll past.”

The company will announce the layoffs on May 20, according to the report.

Meta announced the layoffs earlier than it had intended because details of the plan had already leaked. Bloomberg said Reuters reported on the plan earlier this month.

“We’re doing this as part of our continued effort to run the company more efficiently and to allow us to offset the other investments we’re making,” the memo said, per the report.

Reached by PYMNTS, Meta declined to comment on the report.

Both the Bloomberg report and the earlier Reuters one attributed the layoffs to the company’s effort to offset the investments it has been making in artificial intelligence infrastructure.

PYMNTS reported in March that plans to spend between $115 billion and $135 billion this year as it races to construct data centers, chips and other AI infrastructure. This level of spending puts it in the company of some of the biggest investors in AI infrastructure, including Amazon, Google and Microsoft.

It was reported in January that Meta was cutting 10% of the jobs in its Reality Labs unit, which employed about 15,000 people. That move was characterized as part of a larger plan by the company to reduce its focus on virtual reality products as it concentrates on other AI wearables.

In March, Meta said that over the next few years, it will shift the content enforcement efforts on its apps from the current third-party vendors to the company’s new AI systems.

It was reported in January 2025 that Meta was cutting 5% of its workforce in a move it said would focus on “low performers” and ready the company to build AI, glasses and the future of social media.

In March 2023, Meta said it planned to lay off 10,000 employees after cutting 11,000 jobs in November 2022. The company also said at the time that it would freeze hiring for another 5,000 roles that had been open. When announcing the November 2022 cuts, Meta CEO Mark Zuckerberg said the move was driven by online returning to prior trends after having increased during the pandemic.

Maher Saba, head of the tech giant’s new Applied AI Engineering division, outlined the overhaul in a memo Tuesday (April 14), The Information reported.

Saba, who had been vice president of Reality Labs Foundation before moving to his new role, acknowledged the changes were “big” but argued they were necessary.

“This industry moves fast,” he wrote. “If we don’t adapt, we get left behind.”

The reorganization will see some members of the foundations team move to the Reality Labs wearables group. The quality assurance and “dogfooding” team, as well the trust and platform foundations team, will be embedded directly within the wearables and virtual reality units.

(The report notes that “dogfooding” is a term for testing products and services in-house before a public launch.)

According to the memo, the foundations group will be spread across product teams to allow closer collaboration with the groups they support.

Meta made large cuts to Reality Labs earlier this year, laying off more than 1,500 workers in January and making additional cuts last month.

“Just a few years ago, the metaverse looked like the next platform shift, promising immersive digital worlds where people would live, work and transact at scale,” PYMNTS wrote late last year in a report looking at the company’s Reality Lab struggles.

“Today, Meta’s virtual universe is not dead, but it certainly has shrunk and left is a sobering record of investment, stalled consumer adoption and a set of hard lessons about how innovation in payments and commerce really happens.”

In other Meta hardware news, the company said recently it plans to share what it learns at its Meta Lab stores with the retailers of its AI and virtual reality (VR)-enabled wearables. The tech giant has five Meta Lab stores and uses them to allow shoppers to experience the company’s AI smart glasses and VR headsets.

“We can see what’s drawing people in, and if the layout is confusing, we’ll fix it,” Nicola Mendelsohn, head of global business group at Meta, said last week at Shoptalk, per a Retail Touchpoints report. “We want to bring people together not just to browse but to participate.”

Meta announced in January that it was postponing the global launch of its smartglasses due to “unprecedented demand and inventory” that caused waitlists to extend well into 2026.

“We’ll continue to focus on fulfilling orders in the U.S. while we re-evaluate our approach to international availability,” the company said at the time.

As the Financial Times (FT) reported Monday (April 13), the tech giant has been building photorealistic, AI-powered 3D characters that users can engage with in real time.

Sources told the FT that Meta has recently started to prioritize a Zuckerberg AI character that can interact with workers in his place. One source said the CEO is personally involved in testing and training this likeness, which could provide conversation and feedback to employees.

This involves training the AI on the executive’s mannerisms, tone and publicly available statements, along with his recent thinking on company strategies, so that workers might feel more connected to the digital Zuckerberg through interactions with it.

Sources told the FT that Zuckerberg has become increasingly hands-on as he oversees Meta’s AI efforts, spending five to 10 hours a week coding on various AI projects and sitting in during technical reviews.

PYMNTS has contacted Meta for comment but has not yet gotten a reply.

The FT notes that this AI likeness is separate from Zuckerberg’s efforts to build a “CEO agent” to help him in his job.

A report last month from the Wall Street Journal said this agent has been helping Zuckerberg get information faster, finding answers for him that would normally require going through layers of people. That report added that the project is in keeping with Meta’s view that AI adoption is vital to its success. Zuckerberg addressed this idea on an earnings call in January.

“We’re investing in AI-native tooling so individuals at Meta can get more done. We’re elevating individual contributors and flattening teams,” he said. “If we do this, then I think that we’re going to get a lot more done and I think it’ll be a lot more fun.”

Meta is not alone in creating an AI version of its CEO. During a podcast interview earlier this year, Uber CEO Dara Khosrowshahi said his employees had created an AI clone of their boss. 

Meanwhile, PYMNTS wrote last week about Meta’s new Muse Spark AI model, and the advantage the company enjoys thanks to access to years of user behavior, preferences, and social signals. 

“No other AI company holds that position,” the report said. “OpenAI knows what users have asked previously. Google knows what they search. Meta knows what they buy, who they follow and what they scroll past. It also strengthens Meta’s core business. More context leads to better targeting. Better targeting drives higher-value advertising.”

 

Built over the past nine months under Chief AI Officer Alexandr Wang, the new AI model is showing performance competitive with systems from OpenAI, Google and Anthropic, according to CNBC.

The model now powers Meta’s digital assistant in the standalone Meta AI app and desktop website, with a rollout to Facebook, Instagram, WhatsApp, Messenger and Ray-Ban Meta AI glasses planned for the coming weeks.

Meta spent $14.3 billion to acquire a 49% stake in Scale AI and brought in its founder, Wang, as Meta’s first chief AI officer, after Llama 4, its previous model, fell short of ChatGPT and Anthropic’s Claude.

Multimodal by Design

According to Meta, Muse Spark is a natively multimodal reasoning model that supports tool use, visual chain of thought and multi-agent orchestration.

Most frontier models started as text engines and later bolted on vision. Meta built this model to reason across image, video and text from the ground up, allowing it to act as a virtual assistant and help with people’s daily activities. For example, the model can analyze a video of a user doing push-ups and offer feedback on their form. Muse Spark can also study a photo of a fridge’s contents and suggest dishes for dinner.

Meta worked with over 1,000 physicians to curate training data for more factual and comprehensive health responses, the company said. The model can read a food photo to determine its nutritional content and map the muscles a workout targets.

For harder tasks, a Contemplating mode runs multiple agents in parallel and scored 58% on Humanity’s Last Exam, putting it alongside Google’s Gemini Deep Think and OpenAI’s GPT Pro on complex reasoning.

Meta also flagged gaps in long-horizon agentic tasks and coding. For dinner and pushups, those gaps do not matter. For enterprise software decisions, they do.

The Data Moat No One Else Has

The biggest advantage Meta brings is not the model. It is distribution and data. Logging into the Meta AI app connects a user’s Facebook and Instagram accounts automatically.

As TechCrunch reports, Meta does not explicitly say that personal information from those accounts feeds into the AI, but the company trains on public user data and has positioned Muse Spark as a personal superintelligence product. For anyone who joined Facebook in 2010, that is 15 years of behavior, preferences, and social signals the model draws on.

No other AI company holds that position. OpenAI knows what users have asked previously. Google knows what they search. Meta knows what they buy, who they follow and what they scroll past. It also strengthens Meta’s core business. More context leads to better targeting. Better targeting drives higher-value advertising.

Meta is also embedding commerce directly into the experience. Muse Spark can recommend products, track prices and surface alternatives within social feeds, turning the AI into a shopping assistant across its platforms, according to Axios. Mizuho Securities said usage growth via Shopping mode could drive significant monetization through ad targeting and search.

A Closed Model, a New Revenue Line

With Muse Spark, Meta is moving away from its open-source strategy. Its previous model, Llama, reached millions of developers but generated little direct revenue. This time, Meta is keeping control. Muse Spark is a closed model, and the company has said it will not release its design or code publicly. But Meta said it plans to offer third-party developers access to the underlying technology through an application programming interface.

The company said AI capital expenditures in 2026 will run between $115 billion and $135 billion, nearly twice last year’s figure. That level of spending warrants a return the open-source playbook never produced.

For all PYMNTS AI and digital transformation coverage, subscribe to the daily AI and Digital Transformation Newsletters. 

The firm’s AI-powered Risk Review program has enabled Meta to identify risks earlier, apply safeguards more consistently during product development and monitor the outcomes continuously, according to the post.

The risk review process encompasses the identification and mitigation of potential privacy, safety and security concerns, as well as compliance with applicable legal requirements, across products and features for smartphones, computers and wearable devices, the post said.

At Meta, AI now automates and optimizes parts of this process by prefilling documentation, surfacing relevant product requirements, reducing time spent on intake, helping reviews move faster, and helping teams quickly scan product proposals, per the post.

“Importantly, this AI evolution within Risk Review doesn’t replace human judgment — it strengthens it,” Michel Protti, chief compliance and privacy officer, product at Meta, wrote in the post.

“Now, with the help of AI, people can spot patterns sooner and identify things that may otherwise slip through the cracks. By pairing the efficiency and scalability of AI with the nuance and expertise of humans, we’re delivering better protections for the billions of people who use our products and services every day.”

This announcement follows some other reports released this month about how Meta is deploying AI across its operations.

It was reported March 24 that Meta Chief Technology Officer Andrew Bosworth will lead the company’s efforts to adopt AI tools throughout its workforce. Bosworth said in a post on X that Meta has been integrating AI tools across the organization and expects them to give employees more power to accomplish their work.

On March 23, it was reported that Meta CEO Mark Zuckerberg is building a “CEO agent” to assist him with his job. This AI agent has been helping Zuckerberg get information faster.

Meta said March 19 that over the next few years, it will shift the content enforcement efforts on its apps from the current third-party vendors to the company’s new AI systems. These systems will handle tasks such as repetitive reviews of graphic content, while people will still play key roles in critical decisions.

On March 11, the firm said it launched new AI-powered anti-scam tools for its platforms WhatsApp, Facebook and Messenger. Meta said these tools are designed to help users spot and avoid scammers.

The Instagram Plus subscription is being tested in Mexico, Japan and the Philippines, and possibly other countries, the report said, citing information from users.

Meta told the publication that the company will continue testing the subscription before expanding it to additional countries.

The subscription’s exclusive features include the ability to view a Story without the poster knowing the subscriber viewed it; the ability to see how many people rewatched the subscriber’s own Stories; the ability to customize who sees each Story; and the ability to increase the visibility of the subscriber’s Stories, the report said, citing Meta.

Bloomberg also reported on Meta’s tests Monday, adding that the firm is looking to discover which features are most valuable to Instagram users.

The report said that the company has offered subscriptions of different types in the past and currently offers one that enables users to support content creators.

Meta said in September that it was rolling out a new option in the United Kingdom that offers Instagram and Facebook users the option to subscribe to gain an ad-free experience.

The company said it would continue to offer free access to the two social media platforms, with ads, and would still allow these users to control their ad experience with the tools that were available at the time.

“Subscriptions, as an alternative to seeing personalized advertising, is a well-established and economically viable business model spanning many industries, from news publishing and gaming to music and entertainment,” Meta said in a September press release.

Another social media platform, Snapchat, said in February that it was introducing a subscription program to help creators boost revenues.

This subscription offering lets fans access exclusives such as subscriber-only Snaps and Stories, as well as priority replies to a creator’s public Stories and an “ad-free experience.”

For creators, the Snapchat program offers “the freedom to experiment, reward their biggest supporters, and build a recurring income stream on Snapchat that grows with their audience,” as well as a new and “sustainable source of revenue on top of existing monetization options.”

The company operates five Meta Lab stores and uses them to enable visitors to experience the company’s AI smart glasses and VR headsets, according to the report.

“We can see what’s drawing people in, and if the layout is confusing, we’ll fix it,” Nicola Mendelsohn, head of global business group at Meta, said last week at Shoptalk, per the report. “We want to bring people together not just to browse but to participate.”

Meta Lab stores are currently operating in Honolulu, Hawaii; Burlingame, California; Las Vegas, Nevada; Los Angeles, California; and New York City, according to a Meta web page dedicated to the stores. The site said each location offers immersive demos and interactive exhibits.

It was reported March 18 that Meta signed a 10-year lease on a location on Fifth Avenue in New York City, with Matt Jacobson, vice president and creative director of wearables at Meta, saying the firm was “proud to make a long-term commitment to Fifth Avenue, the heart of U.S. retail.”

Meta opened a Meta Lab pop-location in New York City in November before announcing this permanent location, it said in a Nov. 14 update to a blog post.

When announcing the 10-year lease in a March 18 update to the same post, Jacobson said these locations are designed for hands-on discovery and help visitors see how technology can fit into their everyday lives.

“Our people-first approach to experiential retail is rooted in culture, creativity and self-expression — and it’s driving meaningful sales and awareness of our AI wearables,” Jacobson said. “Placing our flagship store alongside the brands that help define culture, will distinguish Meta Lab from traditional consumer electronics retail.”

Meta announced in January that it pushed back the worldwide launch of its smartglasses due to “unprecedented demand and inventory” that led to waitlists that extended well into 2026.

“We’ll continue to focus on fulfilling orders in the U.S. while we re-evaluate our approach to international availability,” the company said at the time.