“Price is what you pay; value is what you get.” - Warren Buffett
In business, a dollar earned isn’t always just a dollar. The way that dollar is earned can make a world of difference – especially during times of transition. After listening to countless earnings calls over the years and after spending 29 years at Microsoft, I’ve learned first-hand that understanding how a company makes money is as important as how much money it makes. Let’s explore why not all dollars are created equal, and why the quality of revenue matters for strategy and the future of a company.
When $1 from Consulting ≠ $1 from Software
Years ago—before “cloud computing” was even a thing—I learned this lesson early in my Microsoft career. Back then, Microsoft made a healthy profit selling software licenses. Once the software was built (a substantial, upfront sunk cost), each additional license cost virtually nothing to deliver—meaning almost pure profit.
When I joined, I was part of a unit call Microsoft Consulting Services (MCS). Here we tracked billable hours, consultant utilization, and every metric a typical consulting firm cares about. But it didn’t take long for me to realize that those metrics—and the revenue behind them—weren’t really a focus in the greater part Microsoft—outside MCS. For Microsoft at large, Consulting was simply seen as a means to an end, not an end in itself. Also, services revenue were almost never mentioned at the quaterly investor calls. Wall Street barely blinked at consulting revenue; those dollars simply didn’t shine as brightly.
The reason is simple: consulting is a people-intensive business with much lower profit margins and scalability - compared to software. You can’t massively grow consulting revenue without also hiring a lot more consultants, which means costs climb alongside revenue. By contrast, selling software (or subscriptions as we’ll see in a minute) can scale globally without a proportional increase in cost or headcount.
Investors know this. In fact, professional services businesses rarely crack 40% gross margins, whereas software product companies often enjoy gross margins well above 80%. This huge gap in profitability (not to mention the risk of maintain and keeping a large workforce utilized) makes investors value a software license dollar far more than a consulting dollar. In one analysis, simply having a heavier services mix dragged a software company’s gross margin from ~81% down to 68%, acting like a “boat anchor” on profitability. No wonder Microsoft’s leadership (and its shareholders) viewed consulting revenue as relatively less attractive – it carried a high opportunity cost. Not all dollars were equal, it seemed, and the market cared most about the dollars with the biggest scaling potential.
From Software to SaaS: The Cloud Transition
Then in the late 2000s, everything began to change. The industry started shifting from selling one-off software licenses to delivering cloud services and subscriptions. By the early 2010s, it was clear again: a dollar earned from the cloud was not equal to a dollar from a traditional license sale. In fact, a dollar from a perpetual software license started to look more and more like a liability – a sign of yesterday’s business model – whereas a dollar of cloud revenue looked like the bright lights of the future.
Investors and analysts on Microsoft’s earnings calls became laser-focused on the growth of cloud services revenue. And they still are to this day. Revenue from cloud offerings sold via SaaS (Software-as-a-Service), PaaS (Platform-as-a-Service and even IaaS (Infrastructure-as-a-Service) became the metric that moved the stock price up (or down), while legacy license sales were increasingly seen as stagnant. And consulting revenue? Well, it moved even further back in the buss, so to speak.
Why were cloud (“-as-a-service”) dollars so prized? Because they represent a promise of recurring revenue — a customer who subscribes one quarter is very likely to subscribe again the next, creating a steady, predictable cash stream. That predictability and scaling potential command a premium valuation. In fact, public SaaS companies now trade at a median EV/Revenue multiple of about 7.3×, whereas traditional on-premise software firms typically see multiples in the 2.0–3.5× range—meaning each dollar of recurring revenue is worth roughly two to three‐and-a-half times more than a dollar of one-off license revenue in the market today.
Microsoft’s own transformation bore this out. Under Satya Nadella’s cloud-focused strategy, Microsoft’s cloud segment skyrocketed. By the mid-2010s, the company started reporting its “commercial cloud” run-rate, and the growth was phenomenal. Fast forward to today: Microsoft’s Intelligent Cloud division is bringing in over $100 billion annually, roughly double the revenue of the traditional Office software business. Clearly, shifting to a cloud/subscription model (SaaS) proved to be a winning strategy for long-term growth. A cloud dollar not only delivered higher margin in many cases, but also signaled ongoing customer commitment rather than a one-time transaction.
Through this period, the story was consistent: not all revenue dollars are created equal. Dollars from modern, high-margin, scalable streams (like SaaS subscriptions or PaaS metered billing) commanded a premium, while dollars from older models (like perpetual licenses) were viewed as lower quality. And just when everyone got used to the cloud playbook, along came something new to shake things up again: Artificial Intelligence.
Enter AI: Big Bets and New Metrics
In recent years, a massive shift is underway with Artificial Intelligence (AI). Tech companies are now pouring extraordinary sums into AI infrastructure. Just look at 2025: industry-wide, the major cloud players are expected to spend on the order of $300 billion in capital investment to build up AI capabilities and capacity. Microsoft alone announced it plans to invest around $80 billion of that on its AI infrastructure (largely building out data centers) in a roughly one-year period . These are staggering numbers – enough to make any analyst’s eyes widen and ask, “Will all that spending pay off?”
Investors are, as always, keeping a close eye on those dollars spent. The cost side of AI – expensive supercomputers, hordes of GPUs, new data centers – is under scrutiny because it hits current profits. But just like before, not all spending is equal either: how that AI investment is used matters. Is the money going into training AI models (an expensive one-off task), or into supporting AI usage (ongoing services)? In other words, will these billions be a one-time burn (i.e., training), or will they enable a steady stream of revenue-generating activity (aka. inference)?
On the revenue side, history is starting to repeat itself with a new twist. Cloud revenue is still viewed positively, of course – that’s now the baseline. Traditional license revenue is virtually a non-factor for Microsoft’s top line these days. Consulting services remain a relatively small, break-even piece mainly there to support customers. Now the spotlight is turning to AI-driven revenue. Analysts want to see that all the investment in AI is translating into actual revenue growth from AI-powered services. And importantly, they care which AI dollars are coming in. AI revenue from ongoing usage (providing AI services to customers) is far more appealing than AI revenue from one-time projects.
For example, AI dollars coming from inference (where customers use AI models in apps, cloud services, etc.) are the most desirable, because they indicate continuous usage and adoption of AI. This is especially true when those AI services are used by third parties – say, independent software vendors (ISVs) building AI-powered solutions for their own customers on Microsoft’s cloud. Those usage-based dollars are a strong signal of a sustainable and rapidly growing ecosystem. On the other hand, AI revenue from big upfront training jobs – like renting out cloud servers to train someone’s giant AI model – while still welcome, are more one-and-done. They don’t necessarily guarantee a recurring relationship or signal the promise of future ongoing consumption.
Microsoft’s leadership clearly recognizes this distinction. CEO Satya Nadella recently emphasized that the company’s AI revenue growth is “all inference.” Microsoft isn’t simply selling raw GPU compute for others (like Open AI) to train models; in fact, they’re turning away some of that business because demand for inference (ongoing AI usage) is so high. In other words, Microsoft is prioritizing the cloud AI services that drive continuous consumption. That’s a very telling strategy choice. It shows Microsoft would rather have customers using AI-powered platforms (like Azure AI services or GitHub Copilot) regularly, than just cash a one-time check for a big training run. Consistent with the theme, AI dollars from usage (inference) count for more than AI dollars from one-off (AI training) jobs.
The Bottom Line
So, no – a dollar is not simply a dollar, especially not when it comes to how it’s earned. We’ve seen this story play out multiple times: from software licenses vs. consulting, to cloud subscriptions vs. perpetual sales, and now AI usage/inference vs. upfront AI training. The lesson in each case is that the quality and sustainability of the revenue stream profoundly affect how those dollars are valued inside and outside the company. Understanding this is crucial for any business navigating a transition. Whether you’re shifting to a SaaS model or investing in AI capabilities, remember that it’s not just about racking up sales – it’s about earning the right kind of revenue. After all, how a dollar is earned can tell you a lot about the future potential of the next dollar. Recognizing that fact can make all the difference in shaping a successful business strategy for the new era.
“Where your treasure is, there your heart will be also.” - Matthew 6:21 (The Bible)