AWS growth suggests cloud providers may be the first companies converting massive AI investment into dependable recurring revenue.
The AI industry has spent several years promising that enormous infrastructure investments would eventually produce equally enormous returns. Amazon’s latest results provide one of the clearest signs yet that at least part of that promise is becoming real.
According to Amazon’s second-quarter financial results, Amazon Web Services generated $42.2 billion in quarterly revenue, an increase of 37% from the previous year. It was AWS’s fastest growth rate in 18 quarters. AWS operating income reached $16.6 billion, compared with $10.2 billion during the same quarter last year.
Amazon also reported that its AI services and semiconductor businesses had each exceeded annualized revenue run rates of $25 billion. Those figures matter because they show customers are not merely experimenting with AI. They are paying substantial amounts for the computing power, models, storage and chips required to operate it.
Amazon Is Selling the Picks and Shovels
The easiest way to make money during a gold rush is often to sell equipment to the miners.
Amazon does not need every corporate AI project to succeed. AWS earns revenue while companies train models, store data, run agents, test applications and scale workloads. Whether those projects eventually improve the customer’s profits is a separate question.
That distinction explains why Amazon can demonstrate strong AI-related revenue while many businesses still struggle to calculate their own return. The infrastructure provider gets paid when the computing resources are consumed. The customer gets paid only when the resulting system reduces costs, increases output, improves decisions or generates revenue.
The spending required to maintain that position is staggering. Amazon increased its expected 2026 capital expenditures from $200 billion to $220 billion. CEO Andy Jassy said the company still expects to lack enough capacity to satisfy demand. At the same time, trailing 12-month free cash flow fell to an outflow of $7.6 billion, driven largely by increased property and equipment purchases related to AI infrastructure. Reuters reported on the scale of Amazon’s AI infrastructure spending.
Amazon is making the Death Star-sized investment because it sees customer demand waiting on the other side. Most small businesses do not have that luxury.
AI Spending Is Not the Same as AI Value
The Amazon results do not prove that every AI initiative is productive. They prove that businesses are purchasing infrastructure.
That is an important warning for smaller organizations. An AI project can appear active while producing little measurable value. Employees may use several assistants. Departments may subscribe to overlapping tools. API charges may rise. Consultants may build demonstrations. None of this guarantees that the business is operating more efficiently.
Small businesses should measure AI investments against identifiable operational events:
- How much employee time is being recovered?
- How many repeated questions have been eliminated?
- Has onboarding become faster?
- Are errors or rework declining?
- Can routine work continue when a key employee is unavailable?
- Has customer response time improved?
These questions are less exciting than announcing an AI strategy, but they determine whether the investment is creating value or merely generating another monthly invoice.
Start With One Expensive Business Problem
Amazon can build infrastructure years before the revenue arrives. A 50-person company should work in the opposite direction.
Begin with a recurring problem whose cost is already visible. It might be managers repeatedly answering the same questions, estimators searching for prior job information, employees using outdated procedures or new hires depending on one experienced coworker for weeks.
Establish a baseline before introducing AI. Record how often the problem occurs, who is involved, how long it takes and what mistakes or delays result. Then implement one controlled solution and compare the results.
This approach makes the business case easier to defend. It also prevents the company from purchasing a broad platform before it understands what the platform must accomplish.
The same principle applies when evaluating how an AI knowledge hub saves money across an organization. The value does not come from possessing an assistant. It comes from reducing search time, repeated work, interruptions and dependence on individual employees.
Calculate the Entire Cost
Software subscriptions are only one part of an AI project.
The full cost may include cloud usage, implementation, data preparation, document cleanup, security design, employee training, testing and ongoing maintenance. Internal employee time should also be counted. Assigning managers and subject-matter experts to review content is necessary work, but it is not free.
Businesses should also expect usage costs to grow when adoption succeeds. A tool that appears inexpensive during a pilot may become materially more costly when every employee begins using it daily.
That does not make the investment bad. It means the expected operational benefit must grow faster than the cost.
Build the Foundation Before Scaling
Amazon’s results show that the infrastructure layer is becoming commercially mature. The harder problem for most organizations remains internal: their knowledge is scattered, duplicated, outdated or trapped in employee memory.
Adding more computing power does not repair that information. It simply retrieves and processes it faster.
A sustainable implementation therefore requires approved sources, clear ownership, permissions, testing and measurable outcomes. Maisy’s structured knowledge-hub implementation process begins with discovery and information cleanup before AI enablement because reliable answers depend on the quality of the underlying organizational knowledge.
Amazon may have shown where the AI money lands first: cloud infrastructure, chips and computing capacity. The question for every smaller business is where the money returns.
Until that answer can be tied to a specific operational improvement, AI spending remains spending—not strategy.
