The Overlooked Effect That's Hitting AI Companies' Stock Value in 2025
It's no secret that investors have started dumping tech funds, and there's a reason many overlook that explains why AI companies are especially exposed right now, and in 2025.
Shorter AI Chip Lifespans Are Hurting Profits
Companies investing heavily in AI chips to capitalize on the fast-growing AI market face a challenge: depreciation costs that could reduce their future profits.
According to analysts at Barclays, depreciation on AI chips is a hidden cost that many investors overlook in their valuation analyses.
Depreciation is an accounting method that lets companies spread the cost of a capital investment over its useful life. When major tech companies invest billions in GPU chips, those expenditures aren't recorded directly as costs but are instead spread out as depreciation expenses over several years.
Faster Innovation Cycles Create Financial Challenges
The problem is that the useful life of AI GPU chips is often shorter than expected, especially given the rapid innovation cycle in AI chip development. This leads to higher-than-expected depreciation costs, which in turn drag down profits. Barclays predicts these costs will be significant enough to cut profit forecasts for major cloud companies like Alphabet, Amazon, and Meta Platforms by up to 10% as early as next year.
Barclays internet analyst Ross Sandler says AI chip depreciation is the largest expense for these leading companies and that it could become a bigger problem in 2025.
Significant Impact on Stock Valuations
Tech companies spending hundreds of billions of dollars on expensive GPU chips from companies like Nvidia face rapidly rising depreciation costs, especially since Nvidia now launches new products every year. This makes it hard for these companies to extend the useful life of their server investments beyond six years, which they otherwise use as a strategy to soften the depreciation hit to profits.
According to Barclays, Wall Street will significantly underestimate these costs in the coming years. For example, Alphabet is expected to post $28 billion in depreciation costs in 2026, 24% more than current consensus forecasts of $22.6 billion. For Meta Platforms, the gap is even bigger, with Barclays forecasting $30.8 billion compared to Wall Street's $21.0 billion, representing a potential cost increase of 47%.
The Question of Return on AI Investment
Given these rising costs and the fast pace of innovation, the question of return on AI investment is more relevant than ever. Ted Mortonson, tech strategist at Baird, points out that tech companies are now spending over $200 billion on AI infrastructure, with capex up more than 50%. It remains uncertain when this massive wave of investment will actually deliver the expected returns, with the timing of return on invested capital not expected until 2025 or 2026 at the earliest.