Meta Weaves AI into Portfolio, but Struggles with Behemoth

Meta’s (Nasdaq: META) recently announced quarterly results outpaced market expectations that sent the stock climbing 5% in the after-hours trading session. According to a recent report, the global multimodal AI market size is estimated to grow 37% annually over the period 2025 to 2030 from $1.74 billion in 2024. 

Meta’s Financials

Meta’s Q1 revenues grew 16% to $42.31 billion, ahead of the Street’s expected $41.4 billion. Earnings grew 35% to $16.64 billion. Adjusted EPS came in at $6.43, significantly ahead of the market’s forecast of $5.28.

Among key metrics, Daily Active Users grew from 3.35 billion a quarter ago to 3.43 billion, ahead of analyst estimates of 3.39 billion. Its Threads microblogging service now has 350 million monthly users, compared with 320 million in January. The Meta AI digital assistant now has nearly 1 billion monthly users, growing from 700 million monthly users in January.

By segment, Advertising revenues came in at $41.39 billion, ahead of projections of $40.44 billion. Reality Labs hardware division sales fell 6% to $412 billion, falling short of the analyst expectations of $492.7 million. However, the operating loss of $4.2 billion was less than the $4.6 billion expected by the market.

For the second quarter, Meta expects revenues of $42.5-$45.5 billion compared with analyst estimates of $44.03 billion. It did not provide a revenue outlook for the year, but Meta lowered the range of its 2025 expenses from $114-$119 billion to $113-$118 billion. It continues to invest in the business and increased its capital expenditures outlook from $60-$65 billion to $64-$72 billion. The increased capex outlay was due to its continued investments in data centers and hardware to support AI efforts.

Meta’s AI Initiatives

Meta is focused on AI opportunities to deliver improved business messaging through advertising and agents and delivering engaging experiences for its end consumers. It is building its AI with a goal to simplify the advertising process for businesses. It expects that its AI will be able to create content and target the right market based on businesses telling it their objective and budget outlay. Users will not need to design ad content or define target audiences.

Meta’s AI will improve targeting and finding the interested audiences as the AI model keeps learning. Additionally, as AI agents generate content for many businesses, Meta is convinced that advertising will deliver improved measurable business results at scale. Last quarter, it tested its new ads recommendation model for Reels which already saw conversion rates improve by 5%. Besides advertising, Meta is also working on helping businesses with AI that can also act as a business agent for customer support and sales.

Meta is targeting improved engagement through both better recommendations and creating new types of content. The AI-recommendation engine has already resulted in a 7% increase in time spent on Facebook, 6% increase on Instagram, and 35% on Threads. AI is also driving content creation that is helping people produce better content to share. Meta’s AI is now creating not just text but also photos and videos. It is working on making the content more interactive.

Besides integrating Meta AI into its family of apps, Meta recently released the Meta AI standalone app that has been built on Llama 4. The personalized app lets users talk to it to help them discover content and access Assistant. Now people can use natural conversation with Meta AI to get answers while multitasking and doing other things on their device.

The app integrates with other Meta AI features like image generation and editing that can all be done through a voice or text conversation with the AI assistant. This technology will deliver a more natural voice experience trained on conversational dialogue so that the AI generates voice directly instead of reading written responses.

For now, Meta is not monetizing the app, but in the future, it plans to monetize Meta AI both through ad revenue and by charging users for a premium version of the assistant.

Meanwhile, Meta recently announced delays to the release of Behemoth AI model. Behemoth is expected to be Meta’s most powerful LLM that will be used to train its new and smaller models. Originally planned for a release this April, Behemoth was going to be the flagship model in Meta’s Llama 4 series, with the ability to outperform existing models like GPT4.5 and Claude Sonnet 3.7.

The recently released Llama 4 model Scout has a total of 109B parameters with 16 experts, with 17B parameters active at a time and it supports a context length of 10 million tokens. Llama 4 Maverick model brings a total of 400B parameters with an expanded 128 experts, with 17B parameters active at a time and has a context length of 1 million tokens.

Initially, Behemoth was set for release at with 2 trillion parameters and was expected to surpass offerings from OpenAI, Google and Anthropic on certain evaluations. Reports suggest that Meta AI engineers are not convinced about the capabilities in the Behemoth LLM being significant enough to warrant a release. Training difficulties also appear to have delayed effectiveness and launch. The model was expected to be released in April; it has now been delayed to later this year. Some analysts believe that the delay in the release of the new model would mean that Meta could fall behind companies like OpenAI and Google in the AI model race.

Meta’s stock is trading at $642.32 with a market capitalization of $1.62 trillion. It hit a 52-week high of $740.91 in February and a 52-week low of $442.65 in August last year.

Disclosure: All investors should make their own assessments based on their own research, informed interpretations, and risk appetite. This article expresses my own opinions based on my own research of product-market fit, channel execution, and other factors. My primary interest is in product strategy. While this may have bearing on stock movements, my writings tend to focus on long-term implications. The information presented is illustrative and educational, but should not be regarded as a complete analysis nor recommendation to buy or sell the securities mentioned herein. I am not a registered investment adviser and I am not receiving compensation for this article. I am an investor in this company.

Photo Credit: Achin bm from Pixabay