July 17, 2026
BlogDIY Savings: Back to Basics with AzureÂ
The mandate is in: Flat budgets. But the increase in your Enterprise Agreement is not one that flat budgets can absorb. To make matters worse, the gap is compounding, thanks to expiring price protections, the November 2025 elimination of volume discounts, another round of list price increases. Each renewal is harder than the last, especially because what Microsoft granted at the last signing, it will contest at this one.
Closing the gap will require your best negotiation strategy plus a second track running beside it: Do It Yourself (DIY) savings. These are new savings created by how you run your estate, not by what Microsoft concedes.
DIY savings initiatives have three criteria for success:
- They require no permission. No concession letter, no approval, no ask.
- They land in the current budget cycle, not at some future renewal.
- They require few hard dollars to pursue, so the work pays for itself.
This new DIY Savings series is dedicated to finding savings opportunities that are practical, not theoretical. Every opportunity it covers comes from the field, from customers who have already achieved it. The increases you are working to offset come from the licensing side of the Microsoft bill, but the most accessible savings right now sit on the consumption side.
Azure Pays Back Fastest, Even on Ground Already Mapped
Azure runs on a meter, not a calendar. Every other savings play in the Microsoft estate waits on a calendar: A true-up, a renewal, a contract expiry. But there is no true-up to wait for and no anniversary to time with Azure. The meter charges daily, which means optimization pays daily, and a change made this week shows up on next month’s invoice. Nothing in the Microsoft wallet returns results faster.
Savings here are all about the basics of Azure cost optimization: The ongoing work of matching what you run and what you pay for it, to what you actually need. It is familiar ground because you have already cleared it, probably more than once. Every sweep found money, and every sweep landed on the very next invoice.
And because the meter never stops, a dollar recovered is never just one dollar. A structural change repeats its saving on every invoice that follows, turning money found today into cost avoided tomorrow. A license concession expires with the term that granted it. A consumption efficiency has no expiration date at all, so long as the ground it was built on holds still.
But if the basics of Azure cost optimization are such well-worn ground, what new savings could possibly be left to find?
Your Dashboards Are Green. But Your Savings Have Leaked
Not so fast! Did you think the ground really would hold still?
Your savings just leaked because the ground moved under a portfolio built for stable ground. For a decade, the working assumption was that Azure capacity would be there when you needed it, where you needed it. That assumption can no longer be taken for granted. Demand now exceeds supply across workloads and regions, and Microsoft’s own CFO Amy Hood said so out loud: “We are, and have been, short now for many quarters.” A reservation was always a billing construct, never a capacity guarantee.
The instruments churn too. On July 1, Microsoft retired new purchases and renewals of Reserved Instances across a long list of older virtual machine families. When those reservations expire, the usage under them reverts to pay-as-you-go, and with renewals discontinued, auto-renew has nothing left to execute. The catalog can now change between your reviews, and a correctly built portfolio can be stranded by a retirement notice.
But wait! My dashboard is still green!
Here is why nothing turns red. The dashboard grades the portfolio you bought, whether the commitments you hold are being used, and whether your spend sits under them. The dashboard doesn’t reflect whether the portfolio you hold still matches the portfolio you would buy today, on today’s catalog, under today’s constraints. That comparison is against an alternative the dashboard cannot see, so the drift between the two never registers.
That is where the new savings are found: Money leaking quietly behind green dashboards, still present in your current bill. But the ground is not done moving. New leaks now open faster than any sweep can close them. Finding the money is no longer the hard part. Keeping it found is, and that calls for a different way of working. The best-run estates are not failing on a schedule. The playbook is leaking between the sweeps.
Put Tactical Savings on Autopilot
Running this discipline yourself was the right call for a decade, and the results proved it. What changed is the ground. The mechanical layer of optimization now shifts too often to be chased by hand, and a market of providers exists whose whole business is running it continuously, at machine cadence. No one has won that market yet, and the competition works in your favor.
Measure the move against the criteria this series set:
- It requires no permission. You engage a provider on your own authority. Microsoft has no seat at the table, no concession to grant, and no way to claw the results back at renewal.
- It lands in the current budget cycle. The savings that leaked are sitting in the current bill, and recovering them is a reduction against actuals, not a someday avoidance. The automation then keeps capturing them, so what gets recovered stays recovered.
- It requires few hard dollars to pursue. Some providers work on contingency and take a percentage of realized savings. Others already sit inside a managed service contract you hold today, which is leverage you carry into the first conversation. Two questions keep any model honest: how is the savings baseline defined, and do fees apply to total savings or only to savings the provider actually added.
Secure New Strategic Dividends
The bigger fight is arriving now. AI consumption is an economics of its own built on tokens, GPU hours, and agentic workflows that trigger services across providers in a single task. Little of it maps cleanly to the discipline this article just described.
The FinOps Foundation’s 2026 survey found that the most requested tooling capability is granular monitoring of AI spend, and the most requested capability is, by definition, the one practitioners do not have. Mastering it will fall to proven cost engineers, and they will need something the time and focus that the mechanical layer has been consuming. That is what the recovered hours are for.
The second dividend is optionality. Continuous optimization does not create a multicloud strategy. Architecture, skills, and operating muscle do that. What it creates is a clean cost baseline for weighing Azure against AWS and Google Cloud, and for pricing that option before capacity pressure makes it urgent.
The third dividend arrives at the negotiating table. A continuously optimized estate produces something an unoptimized one cannot, specifically, an honest consumption forecast. Any large commitment sized against an unoptimized run rate is not a forecast of need. It is a promise to preserve today’s inefficiency for the length of the term. Optimize first, and the number you take to the table is one you can defend, reducing the risk that shortfall payments claw back the very savings this article just showed you how to capture.
A negotiated concession must be won again and again, at a cost that climbs each time. DIY savings escape that cycle. They require no permission. They’re compounding instead of decaying. And they leave you positioned for the fight that is already arriving. Back to basics. Start with Azure. The meter is running.
DIY Savings Series
Part 1: Back to Basics with Azure (this article)
