States that made ambitious clean energy commitments five years ago are quietly walking them back. The reason is not political will. It is data centers. The infrastructure powering AI models — the kind used to generate marketing copy, answer customer service queries, and train the next generation of large language models — is consuming electricity at a pace that utility planners did not anticipate and cannot currently meet.
Meta committed over $15 billion to CoreWeave for AI compute capacity in a single deal. Data center developers are hitting billion-dollar valuations in funding rounds. Every major hyperscaler is in a land-and-power grab simultaneously. The energy demand from this expansion is outpacing available supply in dozens of markets across the US, and someone has to pay for the infrastructure build-out. Ultimately, that someone is you.
What Is Happening to Cloud Pricing
Cloud infrastructure costs for small and mid-size businesses have been rising 15 to 25 percent year over year, and the trajectory is not flattening. The drivers compound each other. Power costs to run and cool data centers are rising. The capital cost of building new capacity to meet AI demand is enormous. Chip shortages — particularly for the high-bandwidth memory required by AI workloads — are creating hardware bottlenecks that push up the cost of every compute instance on the market.
AWS, Azure, and Google Cloud have all implemented pricing adjustments in the past 18 months. Some increases are visible as rate changes. Others arrive as reduced performance on the same instance tier you have always used, effectively forcing an upgrade to maintain service levels.
For a small business running a handful of cloud workloads, a 20 percent cost increase over two years may feel manageable. For a healthcare practice with a cloud-hosted EHR, a law firm with cloud document management, or a multi-location retailer with cloud-based POS and inventory systems, that same percentage translates to thousands of dollars per year in unplanned spending.
Three Levers to Control Cloud Costs Now
Right-Size Your Instances
Most businesses are running cloud instances that are significantly over-provisioned. When you first moved a workload to the cloud, you probably chose a larger instance to be safe. That caution made sense at the time. But running a 4-vCPU instance at 12 percent average utilization means you are paying for four times the compute you actually use.
Cloud providers publish utilization data. A quarterly audit of your instance sizes against actual usage consistently finds 20 to 40 percent savings opportunities in environments that have never been right-sized.
Use Reserved and Committed-Use Pricing
On-demand pricing is the most expensive way to run cloud workloads. Every major cloud provider offers reserved pricing for workloads you know will run continuously — typically 30 to 50 percent cheaper than on-demand rates. One-year and three-year commitments both generate meaningful savings for predictable workloads like application servers, databases, and file storage.
The catch is that you have to know what you are committing to. Buying a reserved instance for a workload you end up decommissioning does not save money. This is why a clear cloud inventory matters before you lock in commitments.
Eliminate Idle and Orphaned Resources
Cloud environments accumulate waste. Developers provision test environments and forget to tear them down. Migrations leave old storage volumes sitting around at full cost. Snapshots pile up. IP addresses get allocated and never used. Load balancers continue running after the workloads behind them are gone.
In any cloud environment that has been running for more than two years without a dedicated audit, idle resource costs commonly represent 15 to 30 percent of the total bill. That is pure waste with no business value attached to it.
The Energy Angle: Why Your Provider Choice Matters
Not all cloud providers are managing the energy crisis the same way. Microsoft Azure and Google Cloud Platform have made the most credible commitments to renewable energy procurement and carbon-neutral operations. Both have invested in long-term power purchase agreements for wind and solar at the data center level.
This matters beyond sustainability optics. Providers that invest in renewable infrastructure are building long-term energy cost stability into their operations. Providers that rely on grid power from traditional sources are more exposed to the energy price volatility driving cloud cost increases. Over a five-year horizon, provider selection based partly on energy strategy is a financially rational decision, not just a green one.
Hybrid Cloud: The Strategy That Pays Off in a Rising Cost Environment
When cloud costs were low, moving everything to the cloud made financial and operational sense. In a 15 to 25 percent annual cost-increase environment, the math changes. For workloads that run continuously and predictably, on-premises or co-located infrastructure often becomes cost-competitive with cloud hosting at scale.
Hybrid cloud — running some workloads in public cloud and others on owned or co-located infrastructure — is not a legacy approach. It is the rational response to cloud cost inflation for businesses with stable, predictable compute needs. The key is knowing which workloads belong where, and managing both environments with consistent security and operational standards.
What Norvet Does About This
Cloud cost management requires two things most small businesses do not have in-house: the time to audit infrastructure regularly, and the technical fluency to act on what the audit finds. Provisioning changes, reserved instance purchases, resource cleanup, and right-sizing all require access, expertise, and follow-through.
Norvet audits client cloud environments quarterly. We identify idle resources, right-sizing opportunities, and reserved pricing candidates. We implement changes with client approval and track the cost impact so you can see exactly what was saved and why. We also advise on hybrid cloud transitions where on-premises or co-located infrastructure makes financial sense.
The AI data center buildout is not slowing down. Cloud costs will continue rising. The businesses that manage this proactively will spend significantly less than those that absorb the increases passively.
Norvet manages your cloud infrastructure — we find the waste before it hits your bill. Contact us to schedule a cloud cost audit.
Source Attribution
Article content used with permission from The Technology Press and adapted for Norvet MSP publishing.
View source articleNeed help with Cloud?
From cloud migrations to cost optimization, Norvet MSP builds and manages cloud environments that actually work for your business.
