What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Neither AI infrastructure nor AI software has inherently more durable growth. Infrastructure can capture demand for scarce compute through usage and customer commitments, but it must earn back heavy investment in equipment, energy, networks and capacity. Software can sell AI through subscriptions and established workflows, but its growth lasts only if customers adopt, renew and pay enough to cover the cost of delivering those features.
The useful comparison is not simply whose revenue is growing faster. It is whether customer value supports recurring revenue, whether incremental margins hold up as the business scales, and whether the cash generated justifies the investment required.
What counts as AI infrastructure and AI software?
AI infrastructure includes the computing capacity and related services used to train and run AI: cloud capacity, networking, platforms and, in some business models, hardware sales. Revenue may come from usage charges or customer commitments.
AI software includes applications and features that use AI to help customers complete tasks or improve workflows. It may be sold as a subscription, priced per seat or by consumption, or embedded in an existing product.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteThese are business-model categories, not clean reporting labels. One provider can sell infrastructure, platforms and applications. For example, Microsoft Cloud combines Azure with Microsoft 365 Commercial cloud, while Alphabet describes cloud offerings spanning infrastructure, platform services and applications. Alibaba reports cloud AI products and model services. A company’s reported cloud growth therefore cannot automatically be treated as growth in just one layer.
How do the business models differ?
| Measure | AI infrastructure | AI software |
|---|---|---|
| How revenue is earned | Compute, cloud capacity, networking, platforms and sometimes hardware; usage-based charges or customer commitments. | Subscriptions, per-seat or consumption pricing, embedded features and application revenue. |
| Evidence of demand durability | Customer commitments, utilization, renewals and expansion, revenue per unit, customer concentration and capacity lead times. | Paid adoption, retention and renewal, account expansion, revenue per customer, workflow integration and pricing power. |
| Cost exposure | Capital spending, depreciation, energy, equipment, networks and the risk that capacity is underused. | Often less direct asset ownership, but inference and hosting costs can weigh on gross margins. |
| Key growth risk | Capacity arrives ahead of demand, customers are concentrated, equipment ages or operating costs outpace monetization. | Features do not convert to paid use, churn rises, competition weakens pricing or AI undermines the economics of a legacy product. |
| Useful outcome measure | Incremental cash returns and return on invested capital after full infrastructure costs. | Retention and expansion alongside contribution margin after compute and service costs. |
This framework is an analytical way to compare disclosed revenue drivers and risks, not a standardized industry score. A fast-growing revenue line is evidence of demand, but it does not by itself show whether growth will produce durable cash returns.
Rank #2
What do company results show—and what do they not show?
Recent company disclosures illustrate why growth rates need context. The figures below come from different companies, periods and reporting definitions; they are examples, not a like-for-like industry comparison.
| Company-reported figure | What it measures | Why the qualification matters |
|---|---|---|
| Microsoft reported 15% growth in Microsoft 365 Commercial cloud revenue in FY2025. | Commercial cloud revenue associated with Microsoft 365. | Microsoft attributed growth partly to installed-base expansion and average revenue per user. This is one company’s result, not a software-sector growth rate. |
| Microsoft reported 34% growth in Azure and other cloud services revenue in FY2025. | Azure and other cloud services revenue. | Microsoft Cloud gross margin percentage declined slightly in FY2025, partly because of scaling AI infrastructure. Revenue growth and margin growth can move in different directions. |
| Alphabet reported $91.4 billion in capital expenditures for 2025. | Alphabet’s company-wide capital expenditures. | Alphabet said it expected technical infrastructure investment to increase significantly in 2026. It also expects costs including depreciation, energy, equipment and network capacity to rise as AI requires more compute. |
| Microsoft reported Microsoft Cloud revenue of $137.7 billion in FY2024, $168.9 billion in FY2025 and $214.4 billion in FY2026. | Microsoft Cloud revenue across the three fiscal years. | The segment combines cloud and software offerings; this series is not a pure infrastructure or pure software comparison. |
| NVIDIA reported $194 billion in Data Center revenue, up 68% year over year, for FY2026. | NVIDIA’s Data Center segment. | This is a company segment figure, not a measure of total AI infrastructure revenue across the market. |
| Alibaba reported 40% year-over-year growth in Cloud Intelligence Group external revenue for the quarter reported in March 2026. | Alibaba Cloud Intelligence Group external revenue. | Alibaba also reported that AI-related product revenue was 30% of Cloud external revenue for that quarter. These are company-reported figures and definitions, not evidence of sector-wide profitability. |
Alphabet has also warned that AI products may monetize differently from historical offerings and that revenue mix and margin trends may change. Alibaba’s reported growth in AI products and model services indicates demand, but revenue growth alone does not establish profitability. Neither company’s disclosures resolve the broader question of which business model has more durable growth.
Can infrastructure growth last after a major buildout?
It can, if customers continue to use and expand their workloads enough to support the cost of the capacity. Customer commitments can make future demand more visible, but they do not eliminate the risks of high upfront investment, changing utilization, equipment depreciation or operating costs.
In Amazon’s 2025 shareholder letter, CEO Andy Jassy said that a substantial portion of expected AWS 2026 capital expenditures already had customer commitments. He also described short-term free-cash-flow headwinds and wrote: “We are willing to make large capex investments and endure short-term FCF headwinds for the substantial medium to long-term FCF surplus.” That is management’s rationale, not proof that the expected returns have been realized. The decisive evidence will be how capacity utilization, customer expansion and cash returns develop over time.
For an infrastructure business, the question is not only whether demand exists, but whether the provider can serve it economically. If capacity comes online before customer use, costs can rise while revenue lags. If customers are concentrated or workloads shift, commitments may not translate into broadly durable demand. And if energy, equipment, depreciation and network costs climb faster than monetization, revenue growth may not yield attractive incremental returns.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Can software companies make AI a durable source of growth?
Software providers may have an advantage when AI improves a product customers already use: they can offer it to an installed base, fit it into existing workflows and potentially generate subscription or expansion revenue. Microsoft’s FY2025 explanation of Microsoft 365 Commercial growth—installed-base expansion and average revenue per user—illustrates those potential levers, although the reported growth figure is specific to Microsoft.
Recommended Free Tools
The opportunity is not automatic. Customers may try an AI feature without paying for it, may not renew after an initial purchase, or may see too little benefit to justify a higher price. Competition can limit pricing power, and a new AI feature can add inference and support costs. AI may also alter how a legacy product is monetized rather than simply adding revenue to it.
Software durability therefore depends on paid adoption and retention, not just feature launches or usage. A company needs to show that AI strengthens customer value and revenue per account while preserving contribution margin after compute and service costs.
How should you judge which model has more durable growth?
Use the same sequence for each business, while keeping its reporting boundaries and cost structure in view:
- Identify what the reported segment includes. Check whether it combines infrastructure, platforms, software or other products before assigning its growth to one business model.
- Test the quality of demand. For infrastructure, examine commitments alongside utilization, concentration, renewal and expansion. For software, look for paid adoption, retention, account expansion and evidence that AI supports pricing.
- Compare revenue with the costs required to earn it. Infrastructure analysis should account for capital spending, depreciation, energy, equipment and networks. Software analysis should account for inference, hosting and service costs.
- Look for improvement in incremental economics. The core question is whether added revenue produces attractive cash returns after the investment and delivery costs—not whether revenue is growing quickly in isolation.
- Allow for overlap. A provider may earn money from both infrastructure and software, and the economics of one layer can support or constrain the other. Segment labels do not always reveal where the value is captured.
The strongest evidence of durability is a consistent connection between customer value, recurring or expanding revenue, and healthy incremental economics. The available company examples show that infrastructure and software can both grow, while also showing why growth alone cannot settle the comparison. Alphabet has said it generally focuses first on user experience and then on monetization when developing products and services; that is its stated approach, not independent proof of future returns.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
