Gavin Baker: Investors Overreact to Big Tech's AI Spending
Gavin Baker, a prominent hedge fund manager known for his early investment in SpaceX, suggests that the investment community has been overly critical of the significant capital expenditure by major technology companies on artificial intelligence infrastructure. He asserts that a deeper look into factors like enhancing cash flow, the unwavering appetite for AI technologies, and the underpriced nature of existing computing agreements would alleviate much of the current investor apprehension regarding these substantial outlays.
During an appearance on the "Invest Like The Best" podcast, Baker articulated his perspective, emphasizing that the financial performance of AI-centric businesses is already showing an upward trajectory. He pointed to recent quarterly results as early indicators of this acceleration, suggesting that the initial high investment costs are beginning to translate into tangible returns. This implies that the return on investment (ROI) questions frequently posed by analysts and investors will soon be definitively answered by continued robust financial outcomes.
Baker's comments come at a time when many investors, particularly in July, had expressed concerns by divesting from memory and chip stocks. This sell-off was primarily driven by worries that tech giants, such as Meta, were allocating excessive resources to AI development without a clear, immediate path to profitability or a solid return on these infrastructure investments. For instance, Meta's free cash flow experienced a notable decline year-over-year in the second quarter of 2026, dropping by 91% to $784 million. However, Baker counters this by highlighting a 25% surge in Meta's net cash from operational activities during the same period, underscoring what he views as a significant operational improvement at a substantial scale.
Furthermore, Baker underscored the persistent high pricing of Graphics Processing Units (GPUs) as compelling evidence of strong, ongoing demand for AI computing power. He noted that the continued vertical rise in prices for even older GPU models in 2026 was an unforeseen development in 2024 or 2025, indicating a sustained and intense need for these components. He also mentioned that many cloud service providers have yet to fully capitalize on their extensive AI infrastructure due to a large portion of their computing capacity being locked into older, less lucrative contracts established when GPU costs were lower. This situation means that the contracted computing resources are currently trading at a significant discount compared to the prevailing spot market rates, suggesting future revenue growth potential as these contracts are renegotiated or expire.
Baker's analysis encourages a more patient and nuanced evaluation of Big Tech's strategic AI investments. He argues that the market is underestimating the long-term benefits and the speed at which these investments will begin to yield substantial financial returns, urging investors to look beyond short-term fluctuations and focus on the inherent value and burgeoning cash flow generated by AI innovation.