Meta is selling excess compute not as a primary strategy, but because it lacks near-term AI products to utilize its massive capital expenditure. This move is seen as a way to generate ROI while its internal product strategy, aimed at creating a 'personal super intelligence,' has yet to materialize, raising doubts about their overall AI vision.
The availability of compute from Meta and XAI doesn't indicate a market-wide surplus. Instead, it points to a compute allocation problem. Massive capacity is concentrated in the hands of companies that currently lack sufficient internal inference demand for their own models, while other parts of the market remain constrained.
Meta's current AI tools for creators are a significant missed opportunity. Despite possessing granular data on user engagement, the AI provides generic, blog-post-level advice. This failure to create a personalized 'social media copilot' that leverages unique user data represents a major gap in their AI product strategy.
Instead of pursuing a scattered 'super intelligence' strategy, Meta could find more success by focusing on narrow, high-value consumer AI applications. Similar to how the focused Meta Ray-Bans succeeded where the broader Metaverse vision stalled, dominating specific areas like voice or image models within its apps could be a more viable path.
Meta's consideration of acquiring Kalshi, a real-money prediction market, over play-money alternatives indicates a strategic interest in financially incentivized products. This move is highly risky, as it would integrate betting-like features into a platform already under intense global regulatory scrutiny, potentially jeopardizing its core 'golden goose' advertising business.
