Cloud migration pricing reads scarier than it is, mostly because vendors quote the whole project as one number. Priced per workload, it turns legible: lift-and-shift migration runs $3K to $8K per workload, and a five-server small business lands around $15K all-in. Complex environments, the ones with refactoring on the way, compliance requirements, or serious data volumes, run $50K to $250K.
Two numbers in this post deserve more attention than they usually get. AWS MAP credits can cover 25% to 40% of migration cost, and most small businesses never claim them. And the soft costs, things like training, cutover downtime, and team restructuring, get underestimated by 20% to 30% in nearly every self-built budget we’ve reviewed.
We’re gmware, a software development firm in Austin, TX with delivery centers in Bangalore and Mohali, India. Cloud consulting and the migrations that follow are core delivery work for us, so these are numbers we defend on scoping calls every week. Here’s the per-workload menu, the five cost drivers, and the phase plan with gates.
The four numbers that move the bill
| Migration scope | Typical 2026 cost |
|---|---|
| One workload, lift-and-shift | $3K to $8K |
| Five-server small business, full migration | about $15K |
| Complex SMB environment (refactoring, compliance, data) | $50K to $250K |
Cost by migration scope
What cloud migration costs for a small business in 2026
Build the estimate from the bottom: count your workloads, multiply by the $3K to $8K lift-and-shift band, then add a planning-and-landing-zone foundation and a contingency for the soft costs covered below. That arithmetic is why a five-server shop lands near $15K while a 40-workload environment with a database refactor in the middle climbs into six figures.
You’re probably further along than you think, which shrinks scope. 63% of SMB workloads are already cloud-hosted, and the average small business spends about $21K a year on cloud, so for most companies this project is “move the stubborn remainder,” not “move everything.” Timeline for the typical SMB migration: two to six months.
What the per-workload menu actually includes
When we quote $3K to $8K for a workload, that covers the unglamorous full path: dependency mapping for that server or app, replication tooling and a test copy in the cloud, a validation pass against real traffic patterns, the cutover itself, and a rollback plan nobody hopes to use. The spread inside the band is mostly about state. A stateless web server sits at the bottom. A database with replication lag to manage, or an app whose license is tied to hardware, sits at the top.
What the menu price never includes is rework of the application itself. The moment “while we’re moving it, can we also fix it” enters the conversation, you’ve left migration pricing and entered modernization pricing: different multipliers, different timeline, covered in our legacy application modernization cost guide.
The five cost drivers in a migration quote
Five variables explain almost every quote we’ve seen, ours included:
- Workload count. The menu scales linearly until shared services (auth, file storage, networking) complicate the waves.
- Rehost vs refactor mix. Lift-and-shift is the $3K to $8K band; every workload you re-architect mid-flight is a modernization project wearing a migration badge.
- Data gravity and egress. Terabytes leaving a datacenter, or another cloud, cost transfer fees and, worse, calendar time.
- The parallel-running window. Every week old and new infrastructure run together, you pay for both. Decommission speed is a cost lever, not a cleanup chore.
- Consultant rates and team location. Offshore delivery moves this line substantially; it’s the lever we’re built around.
The five drivers behind every quote
Ask any vendor to break their quote against these five. If they can’t, the quote is a guess with a margin on top.
AWS MAP credits, and why nobody claims them
The Migration Acceleration Program is AWS’s subsidy for moving workloads onto its cloud, and the credits can cover 25% to 40% of migration cost. Most small businesses never see a dollar of it, for unglamorous reasons: the program runs through AWS partners, it requires an assessment artifact before migration starts, and nobody at a 30-person company has spare time to learn a vendor funding program. DIY migrations skip the paperwork and quietly forfeit the subsidy.
If you take one action from this post, make it this: before signing any migration SOW, ask the vendor whether MAP funding applies to your project and who files for it. A partner that won’t do that paperwork for you is leaving your money on AWS’s table.
The soft costs that get left out of the budget
The ones that never appear on an invoice. Training, cutover downtime, and the team restructuring that follows a migration get underestimated by 20% to 30% in typical project budgets. In practice that looks like: the cutover weekend that needed three people instead of one, the month where support tickets doubled because nobody knew where the logs moved, the backup jobs that had to be rebuilt for a new storage model.
Parallel running belongs in this bucket too. The old infrastructure bills until someone physically decommissions it, and decommissioning is the task that always slips. It has no champion once the shiny part ships. Budget the soft-cost line explicitly. Pretending it’s zero doesn’t remove it; it just reschedules it as a Q3 surprise.
What you actually save after migrating
Done well, a lot: post-migration infrastructure savings run up to 66%. But “up to” is doing real work in that sentence. The savings come from right-sizing, reserved pricing, and switching off zombie servers, not from the move itself. A lifted-and-shifted environment that nobody optimizes usually costs more than the datacenter did.
The industry numbers back the warning: 28% to 35% of cloud spend is wasted, and organizations without FinOps discipline waste 32% to 40% versus 15% to 20% for those with mature programs.
Where the savings quietly leak
So treat the migration as buying the option to save, and the first-90-days optimization pass as exercising it. If AI workloads are entering your stack, the same discipline applies to a different bill, and we wrote that up in our LLM cost optimization playbook.
AWS or Azure: how much provider choice changes the cost
Less than the marketing implies. List prices sit within 5% to 10% for most compute and storage, so the decision rarely turns on the rate card. The exceptions are stack-specific: Windows-heavy shops can cut licensing 40% to 55% with Azure Hybrid Benefit, and Azure charges nothing for cross-availability-zone transfer where AWS bills $0.01/GB.
AWS vs Azure, where it actually differs
Our rule, after migrations onto both: the cheaper cloud is the one your team understands. Misconfiguration (oversized instances, orphaned storage, chatty cross-zone traffic) costs more than any list-price delta between the two providers ever will.
How long a small business migration takes
Two to six months end-to-end. Here’s the phase plan we run, with the gate that must pass before the next phase starts:
| Phase | Typical duration | Gate before moving on |
|---|---|---|
| Assess and inventory | 2 to 4 weeks | Complete workload list with dependencies; MAP eligibility checked |
| Landing zone and plan | 2 to 4 weeks | Network, identity, and security baseline approved |
| Migrate in waves | 4 to 12 weeks | Each wave passes smoke tests before the next begins |
| Cutover and decommission | 2 to 4 weeks | Legacy infrastructure actually switched off |
| Optimize | First 90 days after cutover | Right-sizing pass done; savings measured against forecast |
The phase plan, with gates
The gate that matters most is the decommission one. Until the old gear is off, you’re paying twice, and “we’ll get to it next sprint” is how a two-month overlap becomes a six-month one.
When migrating is the wrong call
Sometimes the math says stay put, and a vendor should be willing to say so. 86% of CIOs planned to move at least some workloads off public cloud, though only about 8% plan a full exit. The honest reading is that steady, predictable, license-heavy workloads on owned hardware can be cheaper exactly where they are. If your servers are paid for, your load is flat, and nothing is expiring, the business case needs more than “cloud is the future.”
Regulated data changes sequencing rather than the answer. If you handle PHI, the 2027 HIPAA Security Rule changes should shape the migration design from day one, not get bolted on after. And if the application itself is the problem, not the hosting, you’re shopping for modernization, not migration.
How gmware runs SMB migrations
We scope against the five drivers above and quote fixed-price for the assessment phase, so you learn your real number before committing to the program. An Austin-side architect owns the plan and the landing zone design. DevOps and infrastructure engineers in Bangalore and Mohali execute the waves at offshore economics with overlapping US hours. We file the MAP paperwork because knowing it exists is literally our job, and we hold the decommission gate even when it’s unpopular. Especially when it’s unpopular.
“The migration is the easy part. The decommission is where the savings live, and it’s the phase most projects never finish.” (one of our cloud delivery leads)
Tell us what you’re running: server count, data size, anything with a compliance label on it. We’ll come back within 48 hours with a workload-priced estimate and whether MAP credits apply. Start the conversation.