This year, Mark Zuckerberg fully committed to Meta’s AI plan. In 2025, the pressure increases.
When Mark Zuckerberg gave the keynote address at Meta’s Connect event in September, the words “Aut Zuck Aut Nihil” were displayed across the front of his baggy black shirt.
The lines were a play on the Latin phrase “Aut Caesar Aut Nihil,” which means “Either Caesar or nothing,” or more accurately, “All or nothing.” They were shown in a gray font and all caps. For a business that committed all of its resources to its artificial intelligence strategy in 2024, it was a suitable statement.
In April, Meta announced that it would increase its spending by up to $10 billion in 2024 to fund infrastructure upgrades for its AI initiatives. Investors have warmed up to the company’s expensive AI aspirations, despite the news sending shares plummeting as much as 19% that evening. As of Friday’s market close, Meta’s stock price has risen by about 70% this year, having set a record on December 11.
During a call with analysts in October, Zuckerberg stated, “I think we should invest more there because it’s clear that there are a lot of new opportunities to use new AI advances to accelerate our core business that should have strong ROI over the next few years.”
AI’s “positive impact on nearly all aspects of our work,” he said, was crucial to the company’s efforts to revive its online advertising business after it was severely damaged by Apple’s iOS privacy change in 2021. He said that AI is the foundation of Meta’s more recent initiatives, including its experimental Orion augmented reality headset and Ray-Ban Meta smart glasses, which Zuckerberg thinks might be “the next computing platform.”
Zuckerberg’s remarks regarding AI highlight how Meta has made the technology its main focus, which has a direct effect on the company’s operations and may open up new revenue streams in the future. Zuckerberg wants artificial intelligence (AI) integrated into all of the company’s products, unlike its more traditional services like Facebook and Instagram. This is especially important as rivals like OpenAI continue to gain traction with customers.
The GPT family of AI models from OpenAI powers ChatGPT and other apps, whereas the Llama family of AI models from Meta powers the company’s more recent generative AI products, such as the Meta AI personal assistant. Zuckerberg’s main strategy for introducing generative AI technology to its billions of users is that chatbot.
Zuckerberg stated at Connect that “Meta AI is on track to be the most used AI assistant in the world by the end of this year.”
To further increase the effectiveness of its online advertising platform, the company has been progressively launching new generative AI tools for marketers. Additionally, Meta hopes to create a more enterprise-focused division in the next year with the hire last month of Clara Shih, the former CEO of AI at Salesforce, to head a new business AI department.
Because of Meta’s comprehensive approach to AI, analysts believe the company will have greater success in 2025.
In a report dated December 15, Jefferies analysts named Meta one of the “winners” of generative AI for 2025, stating that the company’s enormous user base is “one of the richest surfaces to introduce Gen AI tools.” In a note last week, analysts at Truist Securities stated that the Meta AI digital assistant might compete with Google as “an answer engine for all kinds of queries” and that the social media giant is probably going to do better in 2025, possibly thanks to providing companies with more sophisticated chatbots for customer support.
The Jefferies analysts stated, “We think META has a unique opportunity to introduce Gen AI tools to the nearly 4 billion users & >200M businesses across its family of apps.”
Meta cited earlier data and CEO remarks regarding AI but declined to comment for this piece.
The growing user base of Meta AI:
With Zuckerberg stating in December that the digital assistant “now has nearly 600 million monthly actives,” Meta has been promoting the number of users of Meta AI.
In 2023, he introduced the Meta AI chatbot to compete with rivals’ generative AI chatbots, particularly OpenAI’s ChatGPT. By integrating Meta AI into the search bars of Instagram, Facebook, WhatsApp, and Messenger in April, the corporation elevated it to the forefront of every app in its empire.
However, it’s challenging to directly compare its use to comparable services like ChatGPT or Anthropic’s Claude because the company doesn’t have a standalone Meta AI app, according to David Curry, data editor for analytics firm Business of Apps.
According to Curry, ChatGPT is by far the most popular of these generative AI chatbot programs in terms of monthly active users, followed by Google Gemini, Microsoft Copilot, Claude, and Perplexity.
Based on information he obtained from the Similarweb web-tracking service, Curry said that the Meta AI “standalone website gets less than 10 million views per month, putting it far below the major services (ChatGPT, Gemini, etc.) and even lower than some mid-range players like Anthropic.”
India is now Meta’s “largest market for Meta AI usage,” according to Susan Li, the company’s finance leader, who spoke to analysts in July. Li said that encouraging indicators of engagement and retention on WhatsApp have coincided with that usage.
Sonny Ravan, an Indian music producer based in Pune, is one of those users. According to Ravan, he finds Meta AI, which he uses on WhatsApp, useful for finding out the background of music he likes. He describes it as excellent for preparation and uses it as a tool to find out more about musicians he intends to meet or collaborate with.
With WhatsApp controlling the Indian market for mobile internet communications, 30-year-old Sathish Thiyagarajan, a technical support engineer for marketing tech company GoX.AI, said he is increasingly using Meta AI as a search tool through WhatsApp.
“If my family or friends are saying something to me and I need to look it up, I won’t go to Google,” Thiyagarajan, who lives in Chennai, India, stated. “All I have to do is switch the phone to speaker mode, and Meta AI will start searching right away.”
However, according to Thiyagarajan, he only makes use of Meta AI when he’s on his phone. His go-to AI chatbot when he’s at his computer is ChatGPT from OpenAI.
The integration of Meta AI into WhatsApp’s search features is not universally embraced.
In Kolkata, India, Jawhar Sircar, 72, a retired government official, described WhatsApp’s Meta AI search function as “quite a nuisance.” This is because the Meta AI technology instantly “picks up whatever has been typed” and produces what he calls superfluous search prompt ideas whenever a user pauses while inputting a name in the search-find box.
Because of the Indian government’s ongoing investment in regional computer infrastructure, Sircar said he believes that businesses, technologists, and other professionals who “are getting hooked to AI” are the main users of Meta AI in India.
In an email, Sircar stated, “Although businesses and professionals have begun utilizing AI, the average user has no need, at least not on the Meta platforms.”
Meta’s advertising AI strategy:
Revenue is still largely determined by advertising.
In December, Meta announced that more than 1 million marketers have produced over 15 million ads in a single month using the company’s GenAI capabilities.
About its picture production capabilities, Meta stated at the time, “We estimate that businesses using image generation are seeing a +7% increase in conversions.”
The average small business advertiser is more likely to use Meta’s GenAI tools for more subtle tasks, according to Stacy Reed, an online advertising and Facebook ads consultant, even though people may associate generative AI with the visually striking and occasionally surreal imagery derived from well-known services like Dall-E or Midjourney.
According to Reed, this includes employing AI to generate several headline variations for advertisements, automatically adjusting ad sizes to fit into users’ Facebook and Instagram apps, and moving specific images within ads to improve the effectiveness of promotions.
According to Reed, advertisers who already produce compelling, imaginative copy can ask Meta’s GenAI technologies for “a little bit more” assistance.
“That’s where their AI tools come in handy,” she remarked.
According to Reed, the numerous small marketers she works with aren’t connecting the new capabilities to artificial intelligence. According to Reed, they “believe that Meta is simply improving the way you build ads.”
Celina Guerrero, an independent corporate sales and training consultant, said she finds Meta’s advertising interface complicated and ever-changing, but she utilizes the company’s GenAI capabilities to help her write headlines for her ads.
According to Guerrero, “it is visually overwhelming from a user experience.”
Guerrero stated that she is discussing ways to leverage Meta’s GenAI capabilities for more complex tasks, such as changing the entire in-line copy of her Facebook ad campaign, which is scheduled for January.
Guerrero remarked, “I don’t want my copy to sound like ChatGPT,” alluding to the bland, unoriginal AI-generated content that is widely available online. “I have two choices: either I spend too many hours editing it, or I don’t use the variations.”
According to Jay Pattisall, principal analyst at Forrester, the majority of large corporations and advertising agencies are using more marketing-specific tools for their generative AI-based ad campaigns. According to him, those services are more reliable than Meta’s integrated AI advertising solutions.
Even said, Meta benefits from the launch of basic GenAI capabilities because it and Google control the majority of the digital ad business. According to Maurice Rahmey, CEO of performance marketing company Disruptive Digital and a former Facebook customer manager, Meta’s generative AI capabilities just need “to be good enough to squeeze out more investment” from advertisers.
Even if they only make such minor, gradual adjustments, it’s better for their business since it’s a business of scale, Rahmey stated.
What will happen to Meta’s business strategy next?
According to some experts, Meta may push corporate technology with its Llama family of open-source AI models after hiring Shih from Salesforce in November.
John Hegeman, head of Meta monetization, said in a statement that Llama’s innovations “represent a significant opportunity for businesses to drive more efficiencies and significantly improve the experiences they offer their customers.”
After working at Salesforce from 2006 to 2009, Shih, one of CNBC’s 2024 changemakers, returned to the firm in 2020. In her most recent position at Salesforce, Shih assisted in managing Einstein GPT for Service and Sales, a GenAI tool designed for customer service and sales representatives.
Shih developed a business app during her initial tenure at Salesforce that enabled users to link their Facebook relationships to their Salesforce CRM. In order to help professionals better grasp how to use social networks for business, she wrote a book in 2009 called “The Facebook Era.”
Several former Meta AI and product leaders told CNBC that, given the company’s past failures in developing enterprise software, Shih’s extensive knowledge will be beneficial.
In May, Meta declared that their business communications product, Workplace, would be discontinued by 2026. Additionally, Meta spun out enterprise startup Kustomer in 2023 for an estimated $250 million after purchasing it for roughly $1 billion in 2020.
According to Ralph Schackart, an online equity analyst at investment bank William Blair, the most sensible course of action for Meta would be to build a bigger company around WhatsApp. According to Schackart, WhatsApp might specifically assist companies in developing chatbots for customer support that use Meta’s GenAI.
Regarding Meta’s WhatsApp commercial AI chatbot prospect, Schackart stated, “In the long run, this is going to evolve into customized sales agents, which is an industry worth over $3 trillion.”
