A cloud migration has two budgets: the cloud bill and the engineering labor to move it. Dysnix's breakdown of cloud migration costs puts labor at 40-60% of total spend. The infrastructure bill, compute, storage, egress, is the part most cost calculators already estimate. The part that swings a budget from manageable to blown is who does the migration work, and at what hourly rate.
What cloud migration actually costs in 2026
Published estimates cluster in a wide band. Cloudaware's enterprise cost breakdown puts small workloads at $20,000 to $100,000, with enterprise programs running into the millions once labor, double-run periods, compliance, and integration work are added in. On the small end, Sequentur's cost tally for a 25-user environment prices a full on-premises-to-cloud migration for mixed workloads at $15,000 to $75,000. Scaling up, figures compiled by Appinventiv put the range at $40,000 for a startup to $600,000 or more for an enterprise running mission-critical workloads.
Cloudaware, Sequentur, and Appinventiv all price the infrastructure bill without separating out the people doing the work. Dysnix's data, linked above, puts labor at 40-60% of the total, meaning a $100,000 migration could carry $40,000 to $60,000 in engineering hours alone, before a single server has actually moved.
Why published ranges swing $15,000 to $600,000+
Three variables stack on top of each other to produce that spread.
Workload size sets the floor. A 25-user shop moving a handful of VMs and a database is a different project than a 500-person company migrating a monolith with a decade of undocumented dependencies.
Migration strategy sets the multiplier. A straight rehost, lift the VM, drop it on a new host, costs far less than refactoring an application to use cloud-native services.
Data volume and downtime tolerance set the schedule, and schedule sets the labor bill. A team that can absorb a weekend of downtime moves faster and cheaper than one that needs a zero-downtime cutover with weeks of parallel running, the double-run period that shows up on hidden-cost lists and rarely in the initial quote.
Who does the labor, and what it costs
The 40-60% labor share is the headline number. What it actually costs you depends on who does the work, and that decision has four real options.
In-house engineers (opportunity cost, not free)
Assigning the migration to your existing team feels like the free option, because no new invoice arrives. The cost just doesn't show up as one. Every hour a senior engineer spends on migration work is an hour not spent on the roadmap you already committed to, and migrations pull your most senior people, the ones who understand the legacy dependencies, not your most replaceable ones. If you're staffing a migration with existing headcount, what a senior engineer costs by country is the number to weigh against the opportunity cost of pulling them off everything else.
External consultants and SIs
If in-house skills are thin, the market answer is a consultancy or systems integrator, and Dysnix's data (cited above) puts that at $150-300/hr. For a migration running a few hundred hours, that's tens of thousands of dollars before the underlying infrastructure bill even starts. Consultancies bring migration-specific experience you may not have internally, but that rate is a bill rate, not a disclosed breakdown of what the engineer actually running your migration earns.
Freelance marketplaces
Freelance developer marketplaces sit between consultancy rates and direct hiring. Lemon.io's published rate tiers list mid-level engineers (2-5 years) at $55-85/hr and senior engineers (5-8 years) at $85-140/hr. Lemon.io's terms of use also set a four-week trial period capped at 160 hours, during which you pay the developer's hourly rate plus a flat $2,000 management fee, before you've confirmed the engineer is the right fit for a migration this specific. A DevOps engineer's rate card for 2026 breaks down how these marketplace rates compare across five vendors if you're staffing beyond just the migration itself.
Nearshore engineering teams
HighCircl's rate for senior engineers runs €45-105/hr ($50-115/hr), with a margin capped at 20% and disclosed separately from what the engineer earns. Matching runs 72 hours instead of the multi-week search a direct hire requires, and for engineers based in EU member states, delivery is GDPR-native (Serbia, also in HighCircl's network, sits outside the EU), relevant if the workloads you're migrating carry EU customer data. If cloud migration is the immediate need, vetted cloud engineers in 72 hours is the direct route.
Run the same 300 hours, roughly two senior engineers for a month, a typical size for a mid-sized rehost project, through each priced option and the spread is stark:
| Who does the work | Hourly rate | Margin | Cost for 300 hours |
|---|---|---|---|
| External consultant/SI | $150-300/hr | Not disclosed separately from the bill rate | $45,000-$90,000 |
| Freelance marketplace (Lemon.io) | $85-140/hr (senior tier) | Not itemized on its rate calculator | $25,500-$42,000 |
| HighCircl nearshore | €45-105/hr ($50-115/hr) | 20%, capped, disclosed on top of engineer pay | $15,000-$34,500 |
If your engagement falls inside Lemon.io's 160-hour trial window, add the $2,000 flat management fee on top of the range above.
In-house doesn't appear in that table because there's no invoice to point to. The opportunity cost of pulling a senior engineer off the roadmap for a month is real money even though it never shows up as a line item.
Hidden costs that blow the budget
Rate math is the easy half of the estimate. The harder half is why migrations blow through whatever number was quoted at the start.
future-processing.com's migration cost guide states that 75% of cloud migrations exceed their budgets, with more than a quarter of enterprises seeing overruns above 20%, adding up to over $100 billion in unplanned spending globally across three years. The site attributes those figures to McKinsey & Company but doesn't link a McKinsey source, so treat the numbers as future-processing's citation, not a McKinsey report you can go pull yourself.
Waste compounds the overrun problem after the migration is technically finished. Flexera's 2026 State of the Cloud report found that estimated wasted cloud spend rose to 29%, the first increase in five years. If no one owns FinOps once the migration team moves on to the next project, that waste keeps accruing long after the migration budget line is closed.
The specific line items that cause overruns are familiar to anyone who has run one: a double-run period where old and new environments both cost money at the same time, training for a team that's never operated the new stack, and a compliance validation pass that wasn't scoped because "compliance costs" got budgeted as a vague percentage rather than tied to who's actually running the workload and where.
Cloud migration cost by strategy: rehost, replatform, refactor
The migration strategy you pick changes the per-unit price more than almost anything else. Dysnix's per-unit pricing puts a straight rehost, lift the virtual machine, drop it on new infrastructure with minimal code changes, at roughly $200-500 per VM, though that low upfront cost usually means higher optimization spend afterward, since a lifted VM rarely takes advantage of anything cloud-native. Refactoring a complex application to use managed services and cloud-native architecture runs $10,000 or more per application, a cost that buys lower long-term operating expense in exchange for a much bigger upfront labor bill.
Replatforming sits between the two: enough changes to take advantage of managed databases or container orchestration without a full rewrite. We couldn't find a published per-unit replatform figure, so treat it as closer to refactor cost than rehost cost, and confirm the specific number with whoever's actually doing your migration before you commit to a budget line.
Budgeting a cloud migration: a practical framework
Building your own number means answering four questions before you request a quote from anyone.
What's actually moving? A handful of VMs and a database is a rehost project measured in per-VM costs. A decade-old monolith with tangled dependencies is a refactor project measured in per-application costs, and the two aren't in the same order of magnitude.
Who has the skill in-house, and what's it costing you to keep them on this instead of the roadmap? If the honest answer is nobody, or your best people, that opportunity cost belongs in the budget even though no invoice states it.
Does the workload need a surge team or a standing one? A one-off migration that wraps in a few months is a case for augmenting your team for the length of the project instead of hiring permanently for a job that ends. If this migration is the first of several ongoing multi-cloud projects, the staffing calculation changes.
What happens after cutover? Budget the double-run period explicitly, put someone's name on FinOps ownership before go-live, and scope the compliance validation pass as its own line item rather than a vague percentage tacked onto the total. Migrations that blow their budget are usually the ones where these four questions got answered informally, in a hallway, instead of in the number that went to finance.
Frequently asked questions
How much does cloud migration cost for a small business?
Sequentur's data (cited above) prices a full migration for a mixed-workload, 25-user environment at $15,000 to $75,000. That figure covers the on-premises-to-cloud project itself. It doesn't include the labor cost of whoever runs the migration, which Dysnix's data puts at 40-60% of total spend on top.
What percentage of cloud migration cost is labor?
Dysnix's data (cited above) puts labor at 40-60% of total migration spend. On a $100,000 migration, that's $40,000 to $60,000 in engineering hours before the infrastructure bill is counted separately.
Should you hire in-house, a consultancy, or a nearshore engineer for a cloud migration?
It depends on how long the need lasts. In-house makes sense if you already have engineers who understand the legacy environment and can absorb the opportunity cost of pulling them off other work. A consultancy at $150-300/hr suits a one-off project where you need migration-specific experience you don't have internally. A nearshore engineering team, HighCircl runs €45-105/hr ($50-115/hr) with a disclosed 20% margin, fits a migration surge that's too big for existing headcount but doesn't justify a full consultancy engagement.
How long does a cloud migration take?
Timeline follows the same variables that drive cost: workload size, migration strategy, data volume, and downtime tolerance. A rehost of a handful of VMs runs weeks. A refactor of a legacy monolith with a decade of dependencies runs months, sometimes longer once you add the double-run period and the compliance validation most quotes skip.