Tax Law Is Funding the AI Infrastructure Boom, Not Creating It

Microsoft Corp.'s federal tax expense fell from $14.1 billion to $2.5 billion year-over-year even as revenue surged, illustrating how the AI infrastructure boom is partly financed through accelerated deductions in the 2025 tax law. Lawmakers should require the Treasury Department and the Joint Committee on Taxation to determine whether these tax benefits changed corporate behavior or merely rewarded investments companies were already making.

Microsoft Corp.’s current federal tax expense fell year-over-year https://news.bloombergtax.com/daily-tax-report/microsofts-us-taxes-drop-revenue-jumps-after-trump-cuts from $14.1 billion to $2.5 billion even as its revenue surged, offering a case study of how the artificial intelligence infrastructure boom is being financed partly through accelerated deductions embedded in the massive 2025 tax law. But before anyone claims policy success, lawmakers owe us evidence that these publicly financed tax benefits changed corporate behavior rather than merely rewarded companies for investments they were already racing to make. Congress should require the Treasury Department and the Joint Committee on Taxation to determine as much. It isn’t sufficient to point to rising capital expenditures after a tax cut and declare victory. Policymakers should distinguish between new investment induced by accelerated deductions, investment that would have occurred later but was pulled forward, and investment for which the timing and scale didn’t shift because companies were going to make it regardless. Microsoft and its competitors are spending tens of billions of dollars to keep up with constantly evolving AI technology. That isn’t because Congress recently rediscovered depreciation, but because falling behind in AI could threaten their core businesses. Unlike a typical investment cycle, over the last year, AI infrastructure has become as close to a strategic necessity as one can get. Every major tech company faces a similar reality: If they fail to build enough computing capacity today, they may risk losing market share tomorrow. Given such an environment, tax policies may shift the timing, scale, and location of investments, but it is much harder to argue that they created the underlying investment imperative. Allowing companies to deduct investment costs sooner can reduce the tax penalty on capital, and much of the resulting tax benefit to companies such as Microsoft reflects timing rather than permanent avoidance. But “timing” isn’t synonymous with “free” in terms of public spending. The government is effectively allowing companies to retain cash now in exchange for tax revenue that may arrive later. The inverse holds true as well, as every dollar of accelerated depreciation is a dollar the Treasury can’t use elsewhere in the current budget. Tax expenditures carry opportunity costs just as direct spending does, and the effect is felt in higher deficits, higher taxes elsewhere, or forgone public investment. Microsoft didn’t suddenly discover a tax shelter, and Congress didn’t invent accelerated depreciation for the tax law. The concept has existed in various forms for decades. The tax law merely restored https://www.irs.gov/newsroom/treasury-irs-issue-guidance-on-the-additional-first-year-depreciation-deduction-amended-as-part-of-the-one-big-beautiful-bill 100% bonus depreciation, allowing businesses to recover the full cost of qualifying investments immediately rather than over a period of years. Opponents may frame this as a corporate giveaway; supporters could argue it is simply proper cost recovery. Depreciating an asset over several years can overstate a company’s real taxable income in the current year because it ignores the time value of money. Yet couching accelerated depreciation as just a timing difference is like arguing that receiving an interest-free loan is just a timing difference. The framing is technically correct, but it glosses over the value of timing. All else being equal, you’d always prefer to spend a dollar 10 years from now as against a dollar today. Companies can also kick the tax bill can even further down the road. When a company is rapidly growing and expanding, making continuous investments as it goes, future tax payments can remain perpetually just over the horizon. As each new round comes due, a new round of capital spending can generate another wave of upfront deductions to cut taxes owed in the current fiscal year. The question then is whether taxpayers are receiving commensurate value in return. The traditional case for full expensing is that if the tax code uses stretched depreciation schedules to make investment more expensive than consumption, businesses will invest less than they might otherwise. Allowing firms to recover their costs quickly lowers the after-tax cost of capital. In theory, firms are encouraged to pursue projects that might not otherwise have made good financial sense. But in 2026, Microsoft isn’t building tens of billions of dollars’ worth of data centers because Congress lets it write them off in the first year they’re put into service. It is building them because every major technology company seems to agree that AI will determine the industry’s winners and losers for the next several years. Choosing not to invest was never really an option. If Congress had left depreciation rules unchanged, it is difficult to imagine Microsoft would have simply abandoned Azure’s AI ambitions. Microsoft’s president announced https://blogs.microsoft.com/on-the-issues/2025/01/03/the-golden-opportunity-for-american-ai/ the roughly $80 billion AI data center plan in a blog post on Jan. 3, 2025 — six months before the tax law was signed. The trajectory was publicly stated, presumably budgeted, and well underway before Congress changed the depreciation rules. The political narrative is still beginning to form on the horizon. But correlation is doing a lot of heavy lifting in assuming that soaring capital expenditures prove the tax law is working. The Treasury and the Joint Committee on Taxation should regularly estimate how much investment is genuinely attributable to policies such as accelerated cost recovery, how much was just pulled forward, and how much would have occurred regardless of policy. Until and unless policymakers can answer those questions, any claims that the 2025 tax law “worked” are little more than guesswork. Whether we celebrate AI as the next great engine of economic growth or worry about its effects on labor markets, creative rights, wealth inequality, or the environment, we mustn’t overlook that the public is helping finance the technology. This should be made explicit and quantified even if it proves to be wise policy and a sound investment. But if Congress intends to make taxpayers silent investors in the AI revolution, it should have the honesty to say so. It shouldn’t congratulate itself for “creating” investment that competitive necessity had already made all but unavoidable. Andrew Leahey is an assistant professor of law at Drexel Kline School of Law, where he teaches classes on tax, technology, and regulation. Follow him on Mastodon at @andrew@esq.social. Read More Technically Speaking Learn more about Bloomberg Tax or Log In to keep reading: See Breaking News in Context From research to software to news, find what you need to stay ahead. Already a subscriber? Log in to keep reading or access research tools and resources.