Meta AI Statistics 2026: Users, Revenue & Llama Data

Last Updated: August 1, 2026

Meta AI reached 1.2 billion monthly active users by Q1 2026, up from 1 billion in May 2025, with 63% of interactions on WhatsApp and India as the largest single market at 142 million MAU per Quantumrun's May 2026 Meta statistics analysis. Meta reported Q2 2026 revenue of $60.8 billion, up 28% year over year, in earnings released July 29, 2026, while total expenses surged 55% to $42.0 billion as Meta's $125-145 billion AI capex program compressed margins and pushed free cash flow down to just $784 million.

The Meta AI user count requires careful interpretation before citing it. The 1.2 billion figure includes users who received an AI response inside WhatsApp, Instagram, Facebook search, or Messenger without deliberately opening Meta AI. Presenc AI's Q1 2026 analysis using stricter active-user criteria estimates approximately 640 million monthly active Meta AI users. Both figures are real. They measure fundamentally different levels of user intent. For comparison with ChatGPT's 1.1 billion MAU or Claude's 245 million MAU, the 640 million active-user estimate is more directly comparable.

This guide covers every significant Meta AI statistic for August 2026, including user counts across all surfaces, Q2 2026 earnings data, advertising AI impact, Llama model downloads, Meta Superintelligence Labs, Reality Labs, capex, and competitive context, with every figure linked to a named primary source

🎯 Before you read on - we put together a free 2026 AI Tools Cheat Sheet covering the tools business leaders are actually using right now. Get it instantly when you subscribe to AI Business Weekly.

Get the Free Cheat Sheet

Sources: Meta Q2 2026 earnings via Variety, Investing.com Q2 2026 slides, Quantumrun Meta statistics May 2026, SQ Magazine Meta AI statistics

Meta AI reached 1.2 billion monthly active users by Q1 2026 per Quantumrun's May 2026 analysis, up from 1 billion in May 2025 when Mark Zuckerberg announced the milestone at Meta's annual shareholder meeting, making it the largest AI assistant by user count under the broadest definition of monthly active users.

Meta AI user growth timeline:

Source: Getpanto.ai Meta platforms statistics, Resourcera Meta AI users

The user count methodology that matters most:

The 1.2 billion figure uses Meta's broadest measurement - any user who received an AI-generated response on any Meta surface including WhatsApp search, Instagram search, Facebook feed AI features, and Messenger. This captures passive encounters where a user did not deliberately open Meta AI. Presenc AI's Q1 2026 analysis using stricter active engagement criteria estimates approximately 640 million monthly active Meta AI users - those who deliberately initiated an AI interaction rather than passively receiving an AI-generated result.

WhatsApp is the dominant Meta AI surface at approximately 380 million monthly active users. Instagram search-AI integration reaches approximately 290 million. Facebook search AI reaches approximately 240 million. Messenger Meta AI reaches approximately 110 million. These figures overlap as users of multiple Meta surfaces are counted across each per Presenc AI's Q1 2026 surface breakdown.

The geographic concentration:

India leads all countries at 142 million monthly active Meta AI users per Quantumrun, driven by WhatsApp's dominance as India's primary messaging platform. Brazil, Indonesia, and Mexico together with India account for approximately 42% of all Meta AI MAU per Presenc AI. Meta AI is structurally the dominant AI assistant in markets where WhatsApp is the primary communication platform, regardless of whether those users chose Meta AI deliberately.

Zuckerberg disclosed on the Q2 2026 earnings call that weekly conversations with Meta AI grew approximately 10 times during 2026 per 247 Wall St's earnings preview. This is the clearest signal of genuine intentional engagement growth beyond the passive surface integration numbers.

The standalone Meta AI app:

The standalone Meta AI app attracted an estimated 10 to 20 million early adopters within its first months, available initially in the US, Canada, Australia, and New Zealand per Resourcera's Meta AI user analysis. This figure is the most directly comparable to ChatGPT's standalone app usage - deliberately chosen, not passively encountered.

Platform scale enabling Meta AI distribution:

Total Meta daily active people reached 3.60 billion in June 2026, up from 3.56 billion in March, per Meta's Q2 2026 earnings. Instagram crossed 2 billion daily active users per Zuckerberg's Q2 2026 earnings call. WhatsApp surpassed 3 billion MAU across 180+ countries. Facebook and Instagram have each crossed 3 billion MAU. Threads reached 400 million MAU by Q3 2025. This installed base gives Meta AI embedded distribution on every platform simultaneously without requiring separate user acquisition.

For context on how Meta AI's 1.2 billion MAU compares to ChatGPT's 1.1 billion MAU and Gemini's 900 million MAU, our AI market share 2026 guide covers the full competitive picture with methodology context.

Meta reported Q2 2026 revenue of $60.8 billion, up 28% year over year, on July 29, 2026, beating analyst expectations of $60.17 billion, while net income of $15.8 billion and EPS of $6.18 missed estimates as total expenses surged 55% to $42.0 billion driven by AI infrastructure spending and a $2.4 billion legal charge per Investing.com's Q2 2026 earnings analysis.

Meta revenue trajectory:

Source: Variety Meta Q2 2026, SQ Magazine Meta AI statistics

The Q2 margin compression story:

Revenue grew 28%. Expenses grew 55%. That gap is entirely AI capex. Meta's $125-145 billion full-year 2026 infrastructure commitment required spending only one-sixth of that budget in Q1 - meaning Q2 through Q4 will see dramatically accelerated spending. Free cash flow collapsed to just $784 million in Q2 from significantly higher levels in prior quarters, reflecting the cash demands of data center construction at this scale. Meta stock dropped 9.64% after-hours on July 29 per Investing.com, as investors weighed the revenue beat against the margin compression trajectory.

Q2 2026 one-time charges:

Meta took $2.4 billion in charges related to legal proceedings in Q2 2026 per Variety's earnings report. Additionally, $1.18 billion in severance expenses related to the May 2026 layoff of approximately 8,000 employees as Meta redirected headcount toward AI roles. Both charges suppressed Q2 net income below analyst expectations.

The $14 billion data center joint venture:

One day before Q2 earnings on July 28, Meta and BlackRock announced a $14 billion El Paso data center venture with BlackRock funds owning 80% per 247 Wall St. The structure allows Meta to de-consolidate significant capex from its balance sheet while maintaining access to the compute capacity - a creative solution to the investor concern about free cash flow compression from infrastructure spending.

Meta's advertising ARPU:

Meta achieved $57.03 in annual revenue per user in FY 2025 per Statista figures cited by SociallyIn's Meta statistics. Emarketer projects Meta will surpass Google in global ad revenue by end of 2026 - the first time in the 14 years Emarketer has tracked this market - driven by Meta's 24.1% projected ad revenue growth rate roughly double Google's expected 11.9%.

For the complete AI spending context including Meta's capex in the full enterprise AI infrastructure picture, our AI spending statistics guide covers every major platform.

AI-driven Advantage+ campaigns have surpassed $20 billion in annual run rate, the Generative Ads Recommendation Model delivered a 5% increase in Instagram ad conversions and 3% increase in Facebook Feed conversions, and AI improvements are lifting ad pricing 6% and impressions 18% simultaneously per Meta's Q2 2025 earnings disclosures.

Meta AI advertising impact:

Source: SQ Magazine Meta AI statistics, Bayelsawatch Meta AI statistics

Why advertising is the real Meta AI revenue story:

Meta AI the chatbot generates zero direct revenue. The AI story that actually moves Meta's financial results is advertising AI - the systems that decide which ads to show which users, how to optimize ad creative, and how to improve targeting precision. Advantage+ at $20 billion+ annual run rate is the single largest AI-monetized product at Meta by revenue. The 5% conversion lift on Instagram ads across 4 million advertisers translates to billions in incremental value for Meta's advertiser base, which is the reason advertisers increase their Meta spending and Meta's ad revenue grows 28% in a quarter where the AI chatbot itself generates no subscription revenue.

In conversations with marketing leaders, this distinction matters enormously. When they evaluate Meta as an AI company, they are evaluating the advertising AI, not Meta AI the chatbot. The chatbot is a consumer product and a data signal. The advertising AI is the revenue engine.

For the complete picture of how AI is reshaping digital marketing and advertising, our AI marketing statistics guide covers the full advertising AI landscape.

Meta's Llama open-source AI model family surpassed 1 billion cumulative downloads on Hugging Face as of March 2025 per Meta's official blog announcement, making Llama the most widely distributed open-source AI model family in the world by download count.

Llama model statistics:

Source: SQ Magazine Meta AI statistics, Bayelsawatch Meta AI statistics, DemandSage Meta AI users

The open-source strategy and why it matters:

Meta's decision to open-source Llama is one of the most consequential strategic choices in AI in 2025-2026. 1 billion downloads means Llama is running inside thousands of companies' products, research projects, and developer tools worldwide. Every company running Llama internally is using Meta's research output rather than paying OpenAI or Anthropic per token - which drives developer familiarity with Meta's model architecture, creates ecosystem dependencies, and positions Meta favorably in any future commercial model offerings.

The Llama 4 capability picture:

Llama 4 Scout and Maverick use a Mixture-of-Experts architecture enabling efficiency at scale. Llama 4 Maverick's 10 million token context window is one of the largest among available models, larger than Grok 4.1 Fast's 2M tokens and dramatically larger than Grok 4.5's 500K. Llama 4 Behemoth - the largest model in the family - remains in training as of August 2026. xAI and OpenAI have each claimed various benchmark leads depending on the specific evaluation; the competitive frontier model race produces benchmark rankings that shift with each new release.

For the complete picture of how Llama compares to GPT, Claude, and Gemini model families, our AI chatbots comparison guide covers every platform.

Meta Superintelligence Labs launched in 2026 with the acquisition of the team from Scale AI led by Alexandr Wang, with Muse Spark as its first major proprietary model released in April 2026, triggering a 9% stock jump on launch day per SQ Magazine's Meta AI statistics.

Meta Superintelligence Labs key facts:

The Muse Spark launch and Muse Image controversy:

Muse Spark's April 2026 launch was well-received by markets - the 9% stock jump reflected investor confidence that Meta was developing proprietary AI capability beyond Llama's open-source distribution. Muse Image, which launched shortly after, let users generate images using public Instagram accounts. Within days, widespread criticism from Hollywood talent agencies over the opt-out policy rather than opt-in consent led to the partial suspension of that specific feature per Variety's Q2 2026 earnings coverage.

The strategic shift from open to proprietary:

Meta Superintelligence Labs represents a meaningful strategic evolution. Llama remains open-source. But MSL is building proprietary models that Meta controls and does not release publicly. This dual-track approach - maintaining open-source goodwill through Llama while developing proprietary frontier models through MSL - mirrors how Google maintains TensorFlow as open-source while developing proprietary Gemini models.

Meta raised its 2026 capital expenditure guidance to $125-145 billion, up from the original $115-135 billion guided in January 2026, with $107 billion in new contractual commitments announced in Q1 2026, making Meta's 2026 AI infrastructure commitment one of the three largest capex programs in corporate history alongside Microsoft and Amazon.

Meta AI infrastructure statistics:

Source: SQ Magazine Meta AI statistics, IndMoney Meta Q2 earnings preview, 247 Wall St Q2 earnings preview

The capex-versus-revenue tension that drove the stock drop:

Meta's Q2 2026 stock fell 9.64% after-hours despite a revenue beat. The market concern is not whether revenue is growing - it grew 28%. The concern is that $42 billion in quarterly expenses against $60.8 billion in quarterly revenue leaves $18.8 billion in gross profit before other costs, with free cash flow collapsing to $784 million. If Q3 and Q4 capex "catches up" to the $125-145 billion full-year commitment, the quarters ahead require approximately $85-100 billion more in spending against approximately $120-130 billion in projected revenue. That arithmetic leaves very little room for error and explains why investors demanded a discount despite the revenue beat.

The BlackRock JV as capex management:

Meta and BlackRock's $14 billion El Paso data center announcement the day before earnings - with BlackRock owning 80% - is a balance sheet management tool. By structuring data center ownership as a joint venture where a financial partner holds majority ownership, Meta reduces the amount of capex it must consolidate on its own books. The compute capacity is still available to Meta. The capex burden is shared with BlackRock funds seeking infrastructure yield. This structure will likely be replicated across future Meta data center projects.

The 2035 revenue projection context:

Meta's projection that AI products could generate $460 billion to $1.4 trillion in revenue by 2035 is the internal justification for the current capex program. At $125-145 billion per year in AI infrastructure spending, Meta would invest approximately $1.25-1.45 trillion over the next decade. The lower end of the 2035 revenue projection ($460 billion) would represent less than the cumulative infrastructure investment. The upper end ($1.4 trillion) would represent approximately a 1:1 return over the decade - before operating leverage and margin improvement are factored in. The range is wide enough to be almost meaningless as a specific projection, but it signals Zuckerberg's conviction that the infrastructure investment is justified.

For the complete AI infrastructure investment picture across all major tech companies, our AI spending statistics guide covers the full capex landscape.

Meta's Reality Labs division posted a $4.03 billion operating loss in Q1 2026 on $402 million in revenue, with cumulative losses since 2020 exceeding $75 billion and full-year 2026 losses guided to match 2025's approximately $19 billion per Meta's CFO Susan Li.

Reality Labs statistics:

Source: SQ Magazine Meta AI statistics, IndMoney Meta Q2 preview

The Ray-Ban Meta AI glasses bright spot:

Ray-Ban Meta AI smart glasses sold 7 million pairs in 2025 per EssilorLuxottica earnings - a genuine commercial success in the wearable AI category. The glasses integrate Meta AI voice assistant, camera, and audio without requiring a smartphone to be held. For a consumer hardware product in an early category, 7 million units in one year is meaningful traction. The delay of mixed reality glasses to 2027 reflects Reality Labs refocusing resources toward AI glasses where adoption is real rather than VR headsets where adoption remains niche.

The $75 billion cumulative loss context:

Reality Labs has lost more than $75 billion since 2020. That is more than most major technology companies earn in revenue in a year. Meta's original metaverse thesis - that VR social spaces would become the internet's primary interface - has not materialized at consumer scale. The pivot toward AI glasses and AI assistant integration represents Meta redefining what "the metaverse" means in practice, from VR worlds to AI-enhanced physical reality.

By the broadest MAU definition Meta AI leads all AI assistants at 1.2 billion monthly users, but by revenue per user, deliberate-use engagement, and enterprise adoption it trails significantly behind ChatGPT's $25 billion ARR and Claude's $47 billion ARR despite having more surfaces and more users.

Meta AI competitive comparison:

Source: Quantumrun Meta statistics, AI market share 2026 guide

The fundamental business model difference:

Meta AI generates zero direct subscription revenue. Every dollar Meta earns from AI flows through advertising. ChatGPT generates $25 billion in ARR through subscriptions and API. Claude generates $47 billion in ARR primarily through enterprise API. Meta's AI monetization is advertising optimization, not AI product monetization - and it is working, with $20 billion+ in Advantage+ annual run rate and 28% Q2 revenue growth. But it means Meta AI as a chatbot is not a revenue line. It is a consumer engagement tool and a data signal that feeds the advertising AI.

The emerging market structural advantage:

In markets where WhatsApp is the primary communication platform - India, Brazil, Indonesia, Mexico - Meta AI is the default AI assistant for hundreds of millions of people who have never installed ChatGPT or Claude. This structural distribution advantage is not reflected in revenue because Meta does not charge for it. But it creates training data, brand familiarity, and habitual use at a scale that no other AI company can achieve in those markets without WhatsApp-equivalent distribution.

For the full competitive AI market picture, our AI market share 2026 guide covers every platform's position with proper methodology context.

AI Market Share 2026: ChatGPT, Gemini, Claude and the Battle for AI Dominance
How Meta AI's 1.2 billion MAU fits in the complete competitive landscape with proper methodology context.

AI Marketing Statistics 2026
How Meta's advertising AI compares to the full AI-driven advertising landscape.

AI Spending Statistics 2026
Meta's $125-145 billion capex commitment in the context of all major AI infrastructure spending.

ChatGPT Statistics 2026: Users, Revenue and Market Share
The complete ChatGPT data for direct comparison to Meta AI's scale.

Anthropic Statistics 2026: Revenue, Valuation and Funding
Claude's $47 billion ARR versus Meta AI's advertising revenue model.

AI Adoption Statistics 2026
Enterprise AI adoption rates including the context for why Meta AI has no enterprise seat count.

What Is Meta AI? Complete Guide 2026
The complete Meta AI product guide - every feature, surface, and use case.

AI Statistics 2026: The Complete Data Guide
The master hub for all AI statistics including Meta AI's market position data.

How many users does Meta AI have in 2026?
Meta AI reached 1.2 billion monthly active users by Q1 2026 per Quantumrun's May 2026 analysis, up from 1 billion when Mark Zuckerberg first announced the milestone at Meta's annual shareholder meeting in May 2025. The 1.2 billion figure uses the broadest measurement methodology - any user who received an AI-generated response on any Meta surface including WhatsApp, Instagram search, Facebook feed, and Messenger. Presenc AI's Q1 2026 analysis using stricter active-engagement criteria estimates approximately 640 million monthly active Meta AI users who deliberately initiated an AI interaction. 63% of all Meta AI interactions occur on WhatsApp. India is the largest single market at 142 million MAU. Zuckerberg disclosed on the Q2 2026 call that weekly conversations grew approximately 10 times during 2026. Source: Quantumrun May 2026, Presenc AI May 2026

What is Meta's revenue in 2026?
Meta reported Q2 2026 revenue of $60.8 billion, up 28% year over year, beating analyst expectations of $60.17 billion per Investing.com's Q2 2026 earnings analysis. Q1 2026 revenue was $56.3 billion, up 33% year over year. Full-year 2025 revenue was $200.97 billion, up 22% from $164.50 billion in 2024. Q2 2026 net income of $15.8 billion missed analyst expectations as total expenses surged 55% to $42.0 billion due to AI infrastructure spending, a $2.4 billion legal charge, and $1.18 billion in severance from the May 2026 layoff of approximately 8,000 employees. Free cash flow collapsed to $784 million in Q2 from significantly higher prior levels. Meta stock fell 9.64% after-hours on July 29, 2026. Source: Variety July 29, 2026

How many times has Meta AI been downloaded or how many Llama downloads are there?
Meta's Llama open-source model family surpassed 1 billion cumulative downloads on Hugging Face as of March 2025 per Meta's official blog announcement. Llama is the most widely distributed open-source AI model family in the world by download count. The Meta AI standalone consumer app attracted an estimated 10 to 20 million early adopters in its first months of availability in the US, Canada, Australia, and New Zealand. The Llama 4 family includes Scout and Maverick (available) and Behemoth (in training), using a Mixture-of-Experts architecture with up to 10 million token context windows. Source: SQ Magazine, DemandSage Meta AI

How much is Meta spending on AI in 2026?
Meta raised its full-year 2026 capital expenditure guidance to $125-145 billion, up from the original $115-135 billion announced in January 2026, primarily for AI infrastructure including data centers, GPUs, and power. Full-year 2025 actual capex was $72.22 billion - meaning 2026 capex will be approximately 73-100% higher year over year. Meta announced $107 billion in new contractual commitments in Q1 2026. The day before Q2 earnings, Meta and BlackRock announced a $14 billion El Paso data center joint venture with BlackRock funds owning 80%, helping Meta de-consolidate significant capex from its balance sheet. Total 2026 expenses are guided to $162-169 billion. Source: SQ Magazine, 247 Wall St

What is Meta Superintelligence Labs?
Meta Superintelligence Labs launched in 2026 with the acquisition of the team from Scale AI led by Alexandr Wang. Its mission is building "personal superintelligence for everyone" per Zuckerberg's framing. The first major model from MSL was Muse Spark, released April 2026, which triggered a 9% stock jump on launch day. Muse Image, launched shortly after, was partially suspended due to criticism over its opt-out rather than opt-in consent policy for generating images using public Instagram accounts. Meta Superintelligence Labs represents Meta's first significant proprietary AI research operation separate from the open-source Llama program. Weekly conversations with Meta AI grew approximately 10 times in 2026 per Zuckerberg's Q2 call. Source: SQ Magazine May 2026, Variety Q2 2026

How does Meta AI generate revenue?
Meta AI the chatbot generates zero direct subscription revenue. Meta's AI revenue flows through advertising optimization. AI-driven Advantage+ campaigns have surpassed $20 billion in annual run rate. The Generative Ads Recommendation Model delivered a 5% increase in Instagram ad conversions and 3% increase in Facebook Feed conversions. The Lattice ranking system improved ad quality 12% and lifted conversions 6%. AI is improving Meta's ad pricing by 6% and impressions by 18% simultaneously. More than 4 million advertisers use Meta's generative AI tools. Emarketer projects Meta will surpass Google in global advertising revenue by end of 2026 for the first time in 14 years. The business model is fundamentally different from ChatGPT's $25 billion subscription ARR or Claude's $47 billion enterprise API ARR. Source: SQ Magazine, Bayelsawatch

What are Reality Labs statistics in 2026?
Reality Labs posted a $4.03 billion operating loss in Q1 2026 on $402 million in revenue. Q4 2025 operating loss was $6.02 billion. Cumulative Reality Labs losses since 2020 exceed $75 billion. CFO Susan Li guided full-year 2026 losses to land "on par" with 2025's approximately $19 billion. Ray-Ban Meta AI smart glasses sold 7 million pairs in 2025 per EssilorLuxottica earnings - the clearest Reality Labs commercial success. Mixed reality glasses were delayed to 2027 in December 2025 as Meta redirected Reality Labs spending toward AI glasses where consumer adoption is more established. Source: SQ Magazine, IndMoney preview

Is Meta AI better than ChatGPT in 2026?
Meta AI and ChatGPT serve fundamentally different primary use cases in 2026. Meta AI excels in emerging markets where WhatsApp is the primary communication platform, conversational and discovery queries inside Meta surfaces, and image generation through Meta's creative tools. ChatGPT excels for complex reasoning, coding, document analysis, research, and general-purpose professional tasks. By MAU using broad definitions Meta AI leads at 1.2 billion versus ChatGPT's 1.1 billion. By revenue, ChatGPT's $25 billion ARR dwarfs Meta AI's zero direct subscription revenue. By deliberate standalone usage and professional workflow integration, ChatGPT significantly leads. Neither is universally better. The right platform depends entirely on the use case, the geography, and whether you need a standalone AI tool or an AI assistant inside existing messaging and social apps. Source: Presenc AI Q1 2026

The Meta AI statistics of August 2026 tell a story that requires more context than most headlines provide.

1.2 billion monthly active users is the largest number any AI assistant has ever reported. It is also the number that most requires explanation before citing. 63% of those interactions are on WhatsApp - people who received an AI response in a messaging app that billions of people use daily for reasons completely unrelated to AI. The deliberately-chosen standalone Meta AI app attracted 10 to 20 million users. Both figures are real. They measure profoundly different things.

The revenue story is simpler and more important. Meta's Q2 2026 revenue of $60.8 billion, up 28% year over year, is not driven by AI chatbot subscriptions. It is driven by advertising AI - Advantage+ at $20 billion+ annual run rate, GEM delivering 5% conversion improvements on Instagram, Lattice improving ad quality 12%. That advertising AI is generating real, measurable revenue today. The chatbot is a consumer engagement tool and a signal for the advertising machine.

The capex tension is real and the market reaction on July 29 reflected it accurately. $125-145 billion in 2026 infrastructure spending against $60 billion in quarterly revenue requires either accelerating revenue growth or compressing margins through the investment cycle. The $14 billion BlackRock data center JV announced the day before earnings is Meta's first public attempt to manage the balance sheet strain of that commitment.

The 2035 projection of $460 billion to $1.4 trillion in AI-generated revenue is the anchor for all of it. If Meta is right about AI's role in personal superintelligence, advertising optimization, and business AI over the next decade, the current capex makes sense as infrastructure investment ahead of demand. If the adoption curve is slower than projected or the competitive dynamics shift, the $75 billion Reality Labs precedent is the cautionary analog for what persistent unprofitable investment looks like at Meta scale.

The monitoring metric for 2026: watch the gap between deliberate Meta AI engagement growth (Zuckerberg's 10x weekly conversation figure) and advertising revenue growth (28% in Q2). If AI engagement growth translates to advertising pricing power, the capex is justified. If it stays decoupled, the question of how Meta monetizes its AI assistant beyond ad optimization remains open.