During an interview with Business Insider at the Web Summit tech conference in Lisbon this week, Meta’s Chief Marketing and Analytics Officer Alex Schultz remarked bluntly:
“Obviously, no Meta executive is ever going to answer ‘yes’ to the question of whether we’re overspending on AI.”

According to public plans, Meta is expected to spend up to $72 billion on AI infrastructure this year — a figure that is expected to rise even further next year. Earlier this year, CEO Mark Zuckerberg said he would rather take the “risk of spending tens of billions more” than fall even one step behind in the race toward superintelligence. Meanwhile, Amazon, Google, Microsoft, and private companies like OpenAI are also reaching new historical highs in AI-related capital expenditure, covering chips, data centres, and massive salaries to compete for top AI talent.

Despite the staggering amounts involved, Schultz noted that today’s AI boom is “not outrageous” when measured against total industry market cap or revenue share, especially compared to historical bubbles. He cited the late–19th–century U.S. railroad boom: “Today’s AI boom seems aggressive, but it’s not out of control.”

A Goldman Sachs report released in October estimated that U.S. investment in AI currently accounts for less than 1% of GDP — far below the 2%–5% typically reached during past technological waves such as railroad expansion.

Schultz emphasised that Meta’s AI investments have already translated into billions of dollars of revenue. AI dramatically improves advertising tools and content recommendation algorithms. Meta expects to reach approximately $200 billion in annual revenue by 2025, and its market capitalisation now stands at around $1.5 trillion.

In Schultz’s view, Meta’s most significant AI-driven transformation lies in its more innovative content recommendation system — a necessary upgrade because most user time on Facebook and Instagram has shifted to “unconnected content”, meaning posts not from friends or pages users follow.

He asked rhetorically:
“If we hadn’t made this transition in time, how much would the company have shrunk by now?”
Meta, he said, successfully navigated a significant disruption and generated real business growth.

Schultz also mentioned Vibes, the newly launched AI-generated short-video feed in Meta AI apps. He believes it could become “an important part of the company’s future,” and early data shows strong user retention once people start using it — despite some online critics mocking the feature as “AI slop.”

Energy concerns persist: video-generation models require significantly more compute resources than text or image models, raising questions about electricity consumption and water usage. As video tools like OpenAI’s Sora gain popularity, a broader question emerges: Is their entertainment value worth the environmental cost?

Schultz responded:
“Vibes is nowhere near large enough to drain lakes or require the output of several nuclear plants.”
He added that it is only one of many ongoing experiments to train and refine Meta’s AI models.

He offered a lighthearted cultural critique as well:
“Today’s society — especially in Western Europe — still carries a kind of Calvinist belief that pleasure and fun aren’t legitimate pursuits. But we think life should be enjoyable and interesting; everything else we build is ultimately in service of that.”

Schultz added that the AI wave is prompting valuable societal discussions, including nuclear energy safety and the expansion of freshwater supplies through desalination.
“Overall,” he said, “humanity is capable of creating far more abundance than it has today.”

⭐ Added Commentary / Additional Perspectives

• Meta’s stance reflects the broader dynamic of the AI race: companies fear that under-investing is riskier than overspending. In fields dominated by scale — like AI models, latecomers rarely catch up.

• However, while Schultz dismisses environmental concerns, critics argue that even “small-scale” experiments today can serve as precedents for far more massive rollouts later. The key risk is not current consumption but the future trajectory.

• The shift to unconnected content represents a cultural change: social media once centred on people you know; now it centres on content the system thinks you want. This raises concerns about algorithmic control, user autonomy, and the homogenization of global content.

• Meta positioning entertainment (including AI-generated entertainment) as a core human value is a strategic narrative shift — one that aligns with the company’s future business model but may spark debate about technology’s influence on attention, creativity, and digital well-being.

The above content is compiled by ModeZone, a fashion and entertainment magazine.