Capex Overhang & The Software Valuation Squeeze

The repeated warnings and calls for a slowdown issued by top executives at leading AI labs serve as a timely reminder that the development of state-of-the-art models is far from linear. Despite massive capital commitments, state-of-the-art AI development is facing increasingly formidable physical, regulatory, and power infrastructure bottlenecks. While cumulative capital expenditures across data centers and specialized computing power have reached historic proportions, an accounting reality is beginning to weigh on corporate balance sheets through a widening mismatch between depreciation expenses and near-term monetized revenues.

Massive depreciation charges projected for 2026–2027 are set to compress net profitability and free cash flow across mega-cap technology names, potentially triggering a broader institutional “reality check.” This structural shift occurs as traditional software valuation multiples approach increasingly challenging highs, with price-to-earnings (P/E) ratios often exceeding 45x and price-to-sales (P/S) ratios regularly surpassing 12x. Consequently, this high-valuation backdrop creates an ideal strategic window of opportunity to trim gains from overextended software flagships and rotate capital toward physical AI, industrial automation, and robotics assets that present far more attractive risk-reward entry points.

Physical AI Rotation & ETF Positioning

Unlike pure-play software providers, the robotics and industrial automation segment has recently completed a healthy six- to nine-month consolidation phase, resetting key valuation multiples to highly compelling levels. Among primary investment vehicles, the ROBO ETF offers a exceptionally balanced exposure, trading at a reasonable forward P/E of around 26x, a measured price-to-book (P/B) ratio of approximately 3x, a price-to-cash-flow (P/CF) ratio near 18x, and a price-to-sales (P/S) ratio close to 2x. This valuation discipline stems directly from ROBO’s lower concentration in U.S. mega-cap tech, leaning instead toward premier Japanese and European precision equipment manufacturers. Institutional investors are paying close attention to these competitive moats as capital expenditure shifts toward factory floor deployment.

Similarly, the BOTZ ETF, which maintains higher exposure to computer vision, semiconductor equipment, and medical robotics, presents an attractive decoupling from pure software multiples, trading at a forward P/E between 29x and 32x, a P/B ratio of 3.8x, a P/CF hovering around 20x, and a P/S of roughly 3.5x. Allocating capital into these physical automation leaders allows investors to capture the tangible productivity wave driven by humanoid robotics, autonomous supply chain logistics, and next-generation factory automation. This valuation reset offers a clear margin of safety for long-term investors entering the hardware enablement layer.

Looking ahead, reallocating accumulated profits from overvalued tech software into hardware-centric physical AI provides an optimal tactical entry point over a medium-term horizon. By rebalancing exposure toward high-conviction Chinese and Japanese industrial automation leaders, portfolios gain resilient structural diversification while positioning directly for the next multi-year phase of tangible productivity expansion. As corporate earnings increasingly reward real-world automation over speculative software models, physical AI stands ready to capture a growing share of global enterprise capital expenditures.