Big Tech's AI Revenue Finally Exceeds Data Center Depreciation Costs, Signaling Economic Tipping Point

For two years, the central question hanging over the trillion-dollar artificial intelligence buildout was not whether the technology worked, but whether anyone could actually pay for it. The financial markets have watched with mounting anxiety as the world's largest technology companies poured unprecedented sums into infrastructure, waiting for a sign that the investment would yield sustainable returns. That sign has finally arrived. According to new research from Exponential View, global AI revenue outside of China reached $25 billion in the first quarter of 2026. Crucially, this figure surpassed the industry's estimated $21 billion in quarterly depreciation costs tied to data center and chip investments.

This marks the second consecutive quarter that the artificial intelligence sector has cleared this specific financial hurdle. The milestone provides the first systematic, industry-level evidence that the massive capital spending may be self-sustaining rather than purely speculative. Azeem Azhar, founder of Exponential View, noted that AI demand now consistently clears the depreciation hurdle and appears to be improving over time. Passing this test once could have been dismissed as a statistical anomaly or a temporary spike in enterprise experimentation, but passing it twice establishes a definitive trend that is shifting the narrative on Wall Street.

Understanding the significance of this milestone requires looking at the sheer scale of the ongoing infrastructure buildout. The top technology platforms—Amazon, Microsoft, Alphabet, and Meta—are on track to spend an astonishing $725 billion on capital expenditures in 2026 alone. This represents a 77% year-over-year increase from 2025, marking one of the largest and fastest waves of corporate technology investment in modern economic history. Boards of directors are not approving these historic budgets simply because generative models can write decent emails; they are doing so because the strategic cost of missing the AI infrastructure cycle is viewed as an existential threat to their core businesses.[2]

The depreciation comparison matters because it serves as the most conservative capital-recovery test available to financial analysts. It does not measure pure profitability or return on equity; rather, it asks a fundamental question: are the machines currently running generating enough revenue to eventually replace themselves? Depreciation represents the ongoing accounting cost of the enormous investments that tech companies have made in AI data centers, networking equipment, and advanced graphics processing units (GPUs). By proving that revenue can cover these baseline hardware costs, the industry has established a critical floor for its long-term economic viability.

The revenue validating these investments is materializing through two distinct channels across the technology ecosystem. For cloud infrastructure providers like Amazon and Microsoft, the return on investment is highly direct. It is showing up in enterprise contracts, cloud consumption metrics, and the rapid growth of specialized AI workloads. Microsoft's Azure and Amazon Web Services are seeing their customers scale up spending on both model training and daily inference, directly tying the hyperscalers' massive capital expenditures to tangible, recurring cloud revenue.[2]

For advertising and consumer platforms like Meta and Alphabet, the revenue validation is slightly more indirect but equally lucrative. These companies are proving the value of their AI investments through massive improvements in ad targeting efficiency, heightened user engagement, and vastly superior recommendation systems. Alphabet, which sits uniquely between the cloud and advertising sectors, has seen Google Cloud revenue surge past the $20 billion mark while simultaneously using its proprietary AI models to defend and optimize its core search engine cash flow.[2]

The economic implications of this infrastructure boom extend far beyond the balance sheets of Silicon Valley. The capital expenditure flowing into AI systems—encompassing land acquisition, power contracts, cooling systems, and custom silicon—has become a foundational pillar of the broader economy. Analysts estimate that AI-related capital expenditure is on track to add up to 2.5% to United States GDP growth in 2026. The buildout is so massive that stopping it is no longer merely a corporate policy debate; a sudden halt in AI infrastructure spending would likely act as an immediate recession trigger.

However, financial skeptics are quick to point out that clearing the depreciation hurdle is not synonymous with achieving true profitability. While AI revenue has moved above depreciation costs, those charges still consume more than two-thirds—and by some estimates, up to 84%—of the industry's total AI revenue. This leaves a relatively thin financial buffer to absorb the myriad of other major expenses required to keep the ecosystem running.[1]

Depreciation is, in many ways, the kindest cost line in the artificial intelligence ledger. It completely excludes the staggering operational costs of running modern data centers. To calculate true profitability, companies must factor in the massive electricity bills required to power gigawatt-scale facilities, the advanced liquid cooling systems needed to keep dense server racks from melting, the highly specialized labor force, and the financing costs associated with taking on billions in corporate debt. Clearing depreciation proves the machines are no longer pure promise, but it does not definitively prove the business is throwing off free cash flow.[1]

Furthermore, some industry critics argue that the $25 billion revenue figure may be artificially inflated by circular spending dynamics within the tech sector. A significant portion of the revenue generated by cloud providers comes directly from leading AI labs like OpenAI and Anthropic. These labs, in turn, are heavily funded by the very same hyperscalers they are paying for compute. Critics argue that until a larger share of the revenue comes from non-tech enterprise end-users rather than venture-backed AI startups, the true sustainability of the ecosystem remains somewhat obscured.

Another point of contention revolves around the assumed lifespan of the hardware itself. The Exponential View report assumes a six-year depreciation cycle for AI infrastructure, including the highly coveted GPUs. Some investors worry that this timeline is overly optimistic, given the blistering pace of silicon innovation where new chip generations arrive rapidly and threaten to render older hardware obsolete. However, market data has largely supported the six-year assumption so far, as older generations of chips continue to be rented out at respectable prices for less intensive inference tasks.[1]

Despite the caveats, the transition from speculative infrastructure building to operational "token factories" is undeniably underway. The industry is moving past the phase of merely training massive foundational models and into an era where data centers are running real production workloads at scale. From AI agents automating complex enterprise workflows to generative models accelerating drug discovery candidates, the tangible utility of the technology is finally translating into hard, measurable revenue.

As the major technology companies prepare to report their second-quarter earnings for 2026, the narrative has fundamentally shifted. The bulls and the bears both have better evidence to support their cases, but the burden of proof has eased slightly for the optimists. The $25 billion quarter proves that artificial intelligence is not just a demo, a pilot program, or a founder's ambitious forecast—it is a rapidly maturing commercial engine that is beginning to pay its own way.