Meta's AI Arms Race: Defensive Spending and the Undervalued Potential of Perplexity AI

Meta's $14.3 billion investment in Scale AI in 2024 marked a pivotal moment in its quest to reclaim dominance in the AI race. Yet behind the headlines lies a story of desperation: Meta is scrambling to offset its own internal failures and counter rivals like OpenAI and Google, whose advancements have left its Llama series lagging. While its aggressive acquisitions and talent poaching aim to stem the tide, the company's struggles to retain top researchers and its reliance on costly infrastructure investments suggest its path to leadership is fraught with risk. Meanwhile, Perplexity AI—a startup Meta reportedly tried to acquire—offers a compelling alternative. Its rejection of Meta's overtures signals confidence in its independent growth trajectory, positioning it as a potential undervalued gem in the AI landscape.

Meta's Defensive Play: Buying Time, Not Solutions

Meta's recent moves are less about innovation and more about damage control. The Scale AI deal, for instance, was driven by frustration over its underwhelming Llama 4 release and delays with its Behemoth model. By securing Scale's data-labeling expertise, Meta aims to fast-track its AI training infrastructure—a critical but defensive move. However, this strategy has backfired. Google, Scale's former largest customer, cut ties over data exposure risks, while rivals like Microsoft and xAI followed suit. The exodus underscores a broader problem: Meta's acquisitions often disrupt existing ecosystems rather than strengthen its own.

The Louisiana data center, costing $10 billion, further highlights Meta's defensive spending. While such investments may pay off in the long run, they strain margins in a market where competitors like OpenAI and Google are already scaling efficiently. Meanwhile, Meta's talent hemorrhage—losing 4.3% of its AI team in 2024—suggests its recruitment tactics (e.g., $10M+ annual packages) are insufficient to retain top minds. Of the 14 original Llama researchers, only three remain, with many joining rivals like Mistral AI. This attrition raises a stark question: Is Meta's AI empire built on sand?

The Talent War: Meta's Achilles' Heel

The AI race is a war for talent, and Meta is losing it. While it's luring figures like Scale CEO Alexandr Wang and executives Daniel Gross and Nat Friedman, its attrition rate (second-highest among tech giants) reveals systemic issues. Top researchers cite frustration with Meta's slow decision-making and bureaucratic hurdles. OpenAI and Anthropic, by contrast, offer faster execution and more autonomy—a stark contrast to Meta's corporate culture.

The departure of FAIR director Joelle Pineau and the exodus of Llama team members further illustrate the problem. Without homegrown talent, Meta must rely on acquisitions and external hires—costly and risky bets. In contrast, Perplexity AI has retained its founding team, including CEO Aravind Srinivas, signaling a culture of stability and focus.

Meta's stock performance tells the story: its shares have underperformed Alphabet and Microsoft since 2023, reflecting investor skepticism about its AI strategy. Meanwhile, Perplexity's $14 billion valuation, achieved without a public offering, hints at private-market optimism—a stark contrast to Meta's struggles.

Perplexity AI: The Undervalued Contender

Perplexity's rejection of Meta's acquisition bid in 2023 was a masterstroke. By walking away, it signaled confidence in its independent path—developing advanced search tools and a browser that directly challenges Google's dominance. This strategy aligns with market trends: users crave AI-driven search that's faster and more accurate, and Perplexity's focus on this niche gives it a defensible advantage.

The startup's valuation trajectory is also telling. At $14 billion, it's cheaper than Scale AI (which Meta bought at a 49% stake for $14.3 billion), implying room for growth. Its browser integration and enterprise partnerships (e.g., with Apple in early talks) suggest scalability Meta's walled-garden approach can't match.

Crucially, Perplexity's rejection of Meta's bid wasn't just about pride—it was a strategic move to avoid becoming a cash cow for a fading giant. By staying independent, it can pivot nimbly to new opportunities, such as enterprise AI services or generative tools for developers.

Investment Implications: Play the Underdog

Meta's aggressive spending and talent poaching are defensive measures in a losing battle. Its stock struggles, talent leaks, and reliance on costly infrastructure make it a risky bet. Perplexity, however, represents a compelling alternative. Its valuation suggests it's undervalued relative to its growth potential, especially in a market hungry for differentiated AI tools.

For investors, the path is clear: bet on Perplexity. While it remains private, its valuation trajectory and strategic moves indicate it could IPO or secure another round of funding at a premium. Alternatively, consider ETFs like the Global X AI Development ETF (AID), which includes exposure to AI innovators. Avoid doubling down on Meta unless it can stabilize its talent pipeline and deliver a breakthrough product—neither seems likely anytime soon.

Final Analysis: The Write-Off and the Winner

Meta's AI push is a costly reaction to its declining relevance. Perplexity, by contrast, is a focused disruptor with a clear vision. Its rejection of Meta's bid wasn't just a rejection of a buyer—it was a declaration of independence in a crowded field. For investors, the choice is between a fading giant and an underdog with room to soar. The odds favor the latter.