Tech companies at the forefront of the artificial intelligence revolution face a looming risk that could negatively impact their stock prices. According to Barclays analysts, depreciation costs associated with the massive investments these companies are making in GPU chips could be a significant factor in their future valuations.
Depreciation is an accounting practice that spreads the cost of a fixed asset over its useful life. This allows companies to set aside provisions for replacing the asset when it reaches the end of its lifespan. However, in the case of AI chip investments, Barclays analysts argue that the rapid pace of technological advancements could lead to more frequent upgrades, resulting in higher depreciation costs.
Nvidia, a leading provider of GPUs, has a rapid design cycle, releasing new chips roughly every year. This means that older chips become obsolete quickly, leading to a higher depreciation rate. Barclays tech strategist Ted Mortonson believes that this rapid innovation cycle represents a headwind for AI stocks, potentially impacting their valuations and driving their prices lower over the next year.
Mortonson highlights that Wall Street analysts are significantly underestimating the depreciation costs for companies like Alphabet, Meta, and Amazon. Barclays estimates that Alphabet will face depreciation costs of $28 billion in 2026, while the consensus estimate is 24% lower. For Meta, Barclays estimates $30.8 billion compared to the consensus of $21 billion.
Barclays’ Ross Sandler believes that these discrepancies in depreciation estimates could make Alphabet, Meta, and Amazon shares 5% to 25% more expensive than consensus models. While valuations may not be as stretched as they were during the 2021 bubble, these depreciation disconnects are likely to be scrutinized, especially given the ongoing debate surrounding the valuations of big tech companies.
One way to mitigate the impact of depreciation is to extend the useful life of server assets from five years to six years or more. This would spread the depreciation cost over a longer period, reducing its impact on earnings. However, the rapid pace of innovation in the GPU market could make this measure less effective.
Mortonson points out that AI companies are spending over $200 billion on capital expenditures, which have increased by over 50%. He believes that due to the early stage of this technology and accounting complexities, it will take time to see a return on these investments. He expects that it won’t be until 2025 or 2026 that we see a clear picture of the returns on invested capital.
He emphasizes the need for greater transparency in accounting for the depreciation of GPUs compared to other assets like networking, storage, and servers.
Overall, this analysis suggests that depreciation could be a significant factor in the future performance of AI stocks. Investors need to be aware of these potential costs and the impact they could have on valuations. The rapid pace of innovation in the GPU market adds another layer of complexity, making it crucial for investors to carefully consider the long-term sustainability of these investments.
