Is the Cloud's Growth Hiding Your Compute Waste?

The Question: Are You Wasting Money on Cloud Compute?

In Q2 2026, enterprise spending on cloud infrastructure services hit $143 billion, up 43% year over year (Synergy Q2 2026). That’s the highest growth rate in eight years. But here’s the thing: most of that money is being burned on compute that’s overpriced, underutilized, or just plain wrong for the job. I’ve seen it in every company I’ve consulted for: they pick the easiest option—On-Demand—and never look back. That’s a mistake.

I’m not here to give you a balanced overview. I’m here to tell you that you’re probably overpaying, and I’m going to show you exactly why. The cloud isn’t inherently cheap; it’s only cheap if you make the right choices. And right now, most of you aren’t.

The Real Cost of On-Demand: Convenience at a Premium

On-Demand pricing is the default for a reason: it’s flexible, no commitment, and you pay per second or hour (AWS Documentation). But that flexibility comes at a cost. For steady workloads, it’s the most expensive option. Let’s put some numbers on it.

Take a simple t3.micro instance in us-east-1: $0.0104 per hour, with 2 vCPUs and 1 GiB memory (AWS EC2 Price List API). That’s about $7.50 a month. Sounds cheap, right? But if you run that 24/7 for a year, you’re paying $91. Now, if you commit to a 1-year Savings Plan, you could get up to a 66% discount (AWS Documentation). That’s $31 a year saved—per instance. Multiply that by 50 instances, and you’re throwing away $1,550 a year.

And that’s just the small stuff. For larger instances, the waste is even bigger. A p4d.24xlarge GPU instance costs $21.96 per hour (AWS EC2 Price List API). Run that for a month, and you’re looking at over $15,000. If that GPU sits idle half the time, you’re wasting $7,500. On-Demand is for when you have no idea what you need next week. If you know your baseline, commit.

Spot and Savings Plans: The Cheaper Path

So, what should you do? For steady, predictable workloads, Savings Plans are a no-brainer. They can save you up to 66% on compute (AWS Documentation). For flexible, fault-tolerant workloads, Spot Instances are even better—up to 90% off On-Demand (AWS EC2 Spot). Yes, they can be interrupted with a two-minute warning, but if you design for that, you can slash your compute bill dramatically.

For example, if you’re running CI/CD, big data, or containerized workloads, Spot is perfect (AWS EC2 Spot). I’ve seen teams cut their compute costs by 80% just by moving their CI/CD pipeline to Spot. The two-minute warning is scary, but if your jobs can recover, it’s a no-brainer. And if you’re worried about capacity, use On-Demand Capacity Reservations for the critical bits (AWS EC2 Pricing).

But here’s the kicker: most people don’t even know about the 60% energy savings with Graviton. Graviton-based instances use up to 60% less energy and cost up to 20% less than comparable x86 instances (AWS Graviton). And they’re not just for web servers—Graviton5 has 192 cores and is great for AI workloads (AWS Graviton). So, if you’re not using Graviton, you’re leaving money on the table.

Serverless: The Costly Illusion

Now, let’s talk about serverless. I’ve been harsh on serverless before, and I’ll say it again: it’s not cheaper for steady workloads. Lambda’s free tier is generous—1 million requests and 400,000 GB-seconds per month (AWS Lambda Pricing). But beyond that, you pay $0.20 per million requests and $0.0000166667 per GB-second (AWS Lambda Pricing).

Let’s do the math. If you run a function that uses 1 GB of memory for 1 second, and you invoke it 1 million times a day, that’s 1 million GB-seconds per day. At $0.0000166667 per GB-second, that’s $16.67 per day, or $500 a month. For that same money, you could run a t3.micro for 24/7 for two months. And if you have a constant load, you’re better off with a VM.

But wait, there’s more: Azure Functions has a similar free grant (Azure Functions Pricing), and Google Cloud Functions has 2 million invocations free (Google Cloud Functions Pricing). Yet, the moment you go steady, you’re paying a premium. Serverless is for spiky, unpredictable workloads. If you have a steady load, use a VM.

Containers and Kubernetes: The Hidden Costs

Containers and Kubernetes are all the rage, but they come with their own costs. Amazon EKS is a managed Kubernetes service that automates cluster infrastructure (AWS EKS). But you pay for the control plane, even if you don’t use it. Azure AKS, on the other hand, is free for cluster management; you only pay for the compute nodes (Azure Free Account). That’s a big difference.

And then there’s Fargate, which is serverless for containers. It’s pay-as-you-go, but again, for steady workloads, it’s more expensive than a VM. I’ve seen companies run a simple web app on Fargate and pay 2x what they would on EC2. If you need to scale fast, Fargate is great. If you have a steady load, use EC2 with a Savings Plan.

Comparison Table: On-Demand vs. Savings Plans vs. Spot

As you can see, the choice depends on your workload. But the bottom line is: don’t default to On-Demand.

Bottom Line: The Single Best Move

If you take one thing from this, it’s this: stop using On-Demand for steady workloads. Use Savings Plans for your baseline, Spot for the flexible stuff, and Graviton for everything you can. Start with a small pilot: move one non-critical service to Spot and see the savings. You’ll be surprised.

And if you’re tempted by serverless, remember: it’s for spikes, not for steady streams. Use it where it makes sense, but don’t let it become your default.

Sources

• AWS Documentation - https://docs.aws.amazon.com/ec2/
• AWS EC2 Price List API - https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonEC2/current/us-east-1/index.json
• AWS EC2 Spot - https://aws.amazon.com/ec2/spot/
• AWS Graviton - https://aws.amazon.com/ec2/graviton/
• AWS Lambda Pricing - https://aws.amazon.com/lambda/pricing/
• Azure Functions Pricing - https://azure.microsoft.com/en-us/pricing/details/functions/
• Google Cloud Functions Pricing - https://www.srvrlss.io/provider/google-cloud-functions/
• Synergy Q2 2026 - https://www.srgresearch.com/articles/q2-cloud-market-passes-143-billion-highest-growth-rate-in-eight-years