Azure Catalyst explained: what it changes for SaaS teams
What Microsoft launched and what it changes
Microsoft launched Azure Catalyst on November 17, 2026, as an agent-led cloud migration and modernization service, so it is a migration-planning purchase rather than a SaaS tool your team opens every day. For a small B2B software company it changes something only if you run VMware, Hyper-V, SQL Server, .NET or Java workloads you intend to move to Azure, because the core tools carry no additional charge and the bill comes from the Azure resources, AI services and partner tools the project uses.
Plainly defined, Azure Catalyst is Microsoft's unified agentic modernization solution, and it combines Azure and GitHub Copilot capabilities for discovery, planning, migration and modernization (product page). The launch post, written by Jeremy Winter, Corporate Vice President and Chief Product Officer, Azure Platform, calls it "an intelligent agent-led solution to modernize the apps, data, and infrastructure that already run the business" in its launch post. The same post says the agents modernize "applications, data, and infrastructure in a fraction of the time and cost of traditional methods," and it reports Vendor-claimed count: 1.4 million applications modernized according to Microsoft Azure (2026). The post gives no timeframe for that count and no baseline for the "fraction of the time and cost" comparison, so we treat both as vendor claims.
The rename matters more than the launch copy. Microsoft's product page describes Azure Catalyst as "formerly Azure Migrate" and says existing Azure Migrate projects continue to work in Azure Catalyst (product page). The Microsoft Learn overview repeats the name change and states that the Azure Catalyst agent, formerly the Azure Copilot Migration Agent, is in preview (Learn overview). We did not find independent trade coverage of the rename in this research, so the claim rests on Microsoft's own pages.
On availability, the vendor has not published a general availability date for Azure Catalyst as a whole on the launch post, the product page or the overview we read. The product page says only that "Availability of individual modernization features, agents, and dependent services may vary." Treat the agent as preview and the rest as unconfirmed until Microsoft states otherwise.
TL;DR: Azure Catalyst is the new name and new agent layer on top of Azure Migrate, and its core is free. Whether it matters to you depends on whether you have a legacy estate to move this year.
Key Takeaways
Azure Catalyst is formerly Azure Migrate, and existing Azure Migrate projects keep working, according to Microsoft's product page.
The core tools carry no additional charge; Azure resources, AI services, GitHub Copilot and partner tools are billed separately.
Server replication after the first 180 days is the main line item Microsoft prices directly for servers.
The Azure Catalyst agent is labeled preview on Microsoft Learn, and no overall general availability date has been published.
AWS Transform, the main rival we compare, makes its assessment and several migration agents free and meters its custom transformation agent per agent minute.
Independent analysts are skeptical of AI-led legacy migration; Gartner's 2026 forecast is covered below.
How we evaluated these tools
This is an informed buyer's guide built only from public pages. The weights below are our judgment about what a small B2B software company decides on. They are not vendor data, and no vendor ranked them. We compared Azure Catalyst with the previous version it replaces, Azure Migrate, and with AWS Transform, the closest rival that publishes prices for the same kind of work. We did not evaluate other clouds.
| Criterion | Weight | Why it matters for a small B2B software company |
|---|---|---|
| Fit with your current workloads | 25% | Azure Catalyst names VMware, Hyper-V, SQL Server, .NET and Java; AWS Transform names .NET, mainframe, VMware storage and language upgrades. |
| Price transparency and metering | 20% | Can you forecast spend from published rates before you start? |
| Free scope before paid spend | 15% | Which assessment and migration steps cost nothing? |
| Availability maturity | 15% | Does the vendor state general availability, or only preview, for the parts you need? |
| Target cloud fit | 15% | Where the modernized workload will run, and whether that matches your hosting. |
| Governance and access controls | 10% | Role separation and human approval points before anything moves. |
The weights total 100%. Facts in the tables come from the vendor pages named in the text; the weights and the fit calls are ours.
Feature matrix: Azure Catalyst against AWS Transform
The table below separates what each vendor's pages state from what they leave unstated. Where a page is silent, the cell says so.
| Feature | Azure Catalyst | AWS Transform |
|---|---|---|
| Naming status | Formerly Azure Migrate, per the product page | No renaming described on the pages we opened |
| Workloads named | VMware, Hyper-V, Windows Server, Linux, SQL Server, .NET, Java | .NET and Windows, SQL Server, mainframe, Java, Node.js and Python upgrades |
| Agent status | Azure Catalyst agent in preview, per Microsoft Learn | General availability not stated on the page we opened |
| Where you work | Azure portal, with the agent working in GitHub Copilot to map business logic and dependencies | Kiro, Claude, Cursor and Codex, plus a VSIX IDE extension and an MCP server |
| Free components | Core tools at no additional charge | Assessment, Windows, mainframe and VMware migration agents |
| Metered components | Server replication after 180 days, partner tools, Azure resources | Custom transformation agent per agent minute; continuous modernization |
| Target platform | Azure App Service, Azure Kubernetes Service, Azure SQL and Azure VMs | Amazon EBS, Amazon FSx for NetApp ONTAP and Amazon Bedrock for AI model migration |
| Downstream costs | Azure resources, AI services, GitHub Copilot and partner solutions "may vary" | Resources created from the output are billed at standard AWS pricing |
Compared with the previous version, the continuity is the important part. The migration tooling, including discovery, assessment, replication and Database Migration Service, is carried over, and the new layer is an agent that handles conversational planning and GitHub Copilot-based analysis of business logic. The Learn overview describes the agent as a planning-focused experience that helps you "explore discovered inventory, assess Azure readiness, compare migration strategies, review business case insights" (Microsoft Learn).
What it costs: pricing and TCO
Pricing checked November 17, 2026. Microsoft's pricing page says its figures "are estimates only and are not intended as actual price quotes," and it asks buyers to "contact an Azure sales specialist" for a quote.
Server migration for Azure Catalyst is free for the first 180 days per machine, and after that the fee applies per instance. Server migration after 180 days: $25/month per instance replicated according to Microsoft Azure (2026). The same page says Azure Database Migration Service is free for its first 180 days, and that during the free window you may still pay for underlying Azure Storage, storage transactions and data transfer. Third-party ISV tools are listed with "Get License" links and no prices, so we treat them as Quote-based.
On AWS Transform, the custom transformation agent is metered and the assessment and several migration agents are free. Custom transformation agent: $0.035 per agent minute according to AWS (2026). AWS-managed .NET transformations include up to 50,000 agent minutes at no additional charge, and unused minutes do not carry over to the next month. AWS's own example prices a Java language version upgrade of about 17,000 lines at about 72 agent minutes for $2.52, according to AWS (2026).
For GitHub Copilot, which the Azure Catalyst agent works inside, the plans page lists prices per user. Pro plan, per user: $10 per month according to GitHub (2026). GitHub's Pro+ plan is $39 per user per month and its Max plan is $100 per user per month, according to the same page. The page shows no price for the Business or Enterprise plans, so we treat them as Quote-based.
| Cost item | Vendor | Plan or component | Published price | Our arithmetic (illustrative) |
|---|---|---|---|---|
| Discovery and assessment | Azure Catalyst | Core tools | No additional charge | $0 from Microsoft; Azure resources billed separately |
| Server replication | Azure Catalyst | Per instance, after 180 days | $25 per month | 12 instances × $25 × 6 months = $1,800 |
| Database Migration Service | Azure Catalyst | First 180 days | Free | $0 in the free window |
| Custom transformation | AWS Transform | Per agent minute | $0.035 per minute | 72 minutes × $0.035 = $2.52 |
| .NET transformation | AWS Transform | Up to 50,000 agent minutes | No charge within quota | $0 within quota |
| Seats | GitHub | Pro | $10 per user per month | 12 users × $10 = $120 per month |
| Seats | GitHub | Pro+ | $39 per user per month | 12 users × $39 = $468 per month |
| Partner tools | Azure Catalyst | Third-party ISV tools | Quote-based | Not estimated |
| Seats | GitHub | Business and Enterprise | Quote-based | Not estimated |
Worked example: take a hypothetical estate of 12 VMware servers that a SaaS company replicates to Azure over a 12-month move. The first 180 days are free per machine, and the server fee applies after that, so the billable window is about six months: 12 instances × $25 × 6 months = $1,800 before Azure Storage and data transfer. The Azure Migrate role documentation lists Microsoft.Migrate among the required resource providers, and says that if roles are assigned at resource group scope, project key generation can fail when a provider is not registered at the subscription level, so the subscription owner should register it before the first run (Microsoft Learn).
The table below puts the published figures for the two vendors side by side, using only the rates and example volumes quoted above.
| Published figure | Azure Catalyst | AWS Transform |
|---|---|---|
| Free window or quota | 180 days per machine | 50,000 agent minutes (.NET) |
| Metered rate | $25 per month per instance | $0.035 per agent minute |
| Example volume | 12 instances for 6 billable months | 72 agent minutes |
| Example cost | $1,800 | $2.52 |
Vendor profiles: where each one fits
Azure Catalyst (formerly Azure Migrate)
Best fit: a team that already runs VMware, Hyper-V or SQL Server workloads, plans to move them to Azure, and wants discovery, assessment and migration in one portal. Microsoft's FAQ names VMware, Hyper-V, Windows Server, Linux, SQL Server, .NET and Java as supported scenarios (product page).
Limitations: general availability is not published, the agent is in preview, and the pages we opened do not list mainframe workloads for Azure Catalyst. The launch post says that some previews cover "modernizing applications and databases together" and "migrating from BizTalk to Azure Logic Apps," which are separate preview scopes you should confirm before you rely on them (launch post).
Implementation: Microsoft's product page says to start "by creating or accessing an Azure account and launching the Azure Catalyst agent." Before a project runs, the built-in roles matter. Azure Migrate Owner can create projects, Decide and Plan Expert covers discovery and assessment, and Execute Expert covers replication and migration, which gives you a human approval split if you assign them separately (Microsoft Learn). The same page notes that registering the Azure Site Recovery replication appliance needs the Decide and Plan Expert role at subscription scope, not resource group scope.
AWS Transform
Best fit: teams with Windows and .NET estates, mainframe work, or a steady stream of language-version upgrades, and teams that want to run agents from an IDE. AWS describes it as an agentic AI service that "uses agentic AI to accelerate and de-risk each stage of that journey" (AWS Transform).
Limitations: the page we opened does not state general availability, the custom transformation agent is metered, security analyses are billed by AWS Security Agent, and resources built from the output are billed at standard AWS rates (AWS pricing). AWS's page also makes speed claims, including mainframe work moving "from years to months," and gives no methodology for them, so treat those as vendor claims.
Implementation: the Visual Studio extension is free, and the agents are reachable from Kiro, Claude, Cursor and Codex, per the AWS Transform page. Before you commit spend, decide which agent runs, cap the monthly minutes you will allow, and confirm which downstream AWS resources the output will create.
| Your situation | Better starting point | Why |
|---|---|---|
| Your estate is VMware, Hyper-V or SQL Server on-premises | Azure Catalyst | Those scenarios are named on its product page. |
| Your code is .NET or Java and you already run on Azure | Azure Catalyst | Its planning agent sits in the same Azure portal. |
| Your main job is Windows, mainframe or language-version upgrades | AWS Transform | Those agents are named on AWS's pages, and several are free. |
| You have no legacy estate to move | Neither this month | Neither tool addresses a pure SaaS workload. |
Implementation checklist and review points
Here is the decision checklist we would use before the first run, whichever vendor you pick:
Inventory the workloads you actually plan to move, not the ones on the launch copy.
Check that each workload type is named on the vendor's own supported-scenarios page.
Confirm the role split in Azure RBAC so the person approving a wave is not the person who replicates it.
Set a monthly budget for metered agent minutes or server replication before the first run.
Write down which parts are preview and do not commit a migration date that depends on them.
A proposed US Tech Automations workflow could sit on top of this checklist. This is a configurable design, not a deployed one, and we have not verified every prerequisite in this research. Trigger: a new assessment report export is saved to a shared folder. Action: the workflow reads the export, flags any workload whose estimated Azure cost exceeds a threshold that the buyer sets, and opens one review ticket per flagged workload. Output: a review queue for the engineering lead. Prerequisites: the export must be available to the workflow, which requires the assessment-export permission that the roles page assigns to Owner, Decide and Plan Expert and Execute Expert, and we have not verified the export's schema or whether it carries an estimated-cost field. Human review points: an engineering lead approves any wave before replication, and the workflow never starts replication on its own.
A second proposed workflow would cover AWS Transform spend. Trigger: the first business day of each month. Action: the workflow pulls agent-minute usage from billing data, multiplies the minutes by the per-minute rate on the AWS pricing page, and compares the total with the budget you set. Output: a monthly cost line and an alert when the budget is near. Prerequisite: billing data access that includes AWS Transform usage, which we have not confirmed. Human review point: finance or operations approves any new custom transformation run that would exceed the alert.
The no-code alternative is to build the same flow in Zapier, Make or n8n. Those tools can be configured for run histories, retries, error branches and audit evidence. n8n's error-handling documentation describes an error workflow that runs when an execution fails, an execution history per workflow, and log streaming to external systems, though the page we read does not describe per-node retry settings. Whichever route you choose, you own observability, idempotency, escalation, access controls and maintenance. The proposed design above is a way to configure those review points; it does not remove the need to own them.
When NOT to use US Tech Automations: if the job is one application moved by one engineer, the vendor's own assessment and replication tools are simpler and cheaper to run. If your only need is a two-step notification that Zapier or n8n already handles, a dedicated workflow adds nothing. If you have no legacy estate, neither the Azure Catalyst nor the AWS Transform path is relevant, and a notification or review workflow is the only part worth designing.
Signal vs Speculation
Signal, meaning facts from the pages we opened:
Azure Catalyst is formerly Azure Migrate, per the product page.
The core tools are free, and server replication costs $25 per month per instance after 180 days, per the pricing page.
The Azure Catalyst agent is in preview, per the Learn overview.
AWS Transform's custom transformation agent is metered at $0.035 per agent minute, per AWS's pricing page.
Gartner's forecast is that more than 70 percent of mainframe exit projects started in 2026 will fail to produce the intended benefits, according to The Register. The same article quotes Gartner on "significant limitations when it comes to the automated conversion and migration of legacy code."
Speculation, meaning our judgment:
The rename means existing Azure Migrate users will see the least disruption, and new buyers will mostly notice the agent layer.
The agent features are where Microsoft is putting its effort, so we expect them to change faster than the migration tooling. That expectation is ours, not a vendor commitment.
A small SaaS team with no legacy estate gets little from either tool this month. If you want background on the agent concept, our agentic AI explainer covers it.
AI-led modernization will hit the limits Gartner describes on complex legacy code, so budget for human review rather than assume the agent output is final.
Who this is for
This comparison is for a founder or operations or engineering lead at a small B2B software company that runs VMware, Hyper-V, SQL Server, .NET or Java workloads, or already uses Azure Migrate, and needs to decide whether to act now or wait. It is not for a team whose product runs entirely on managed SaaS with no migration project on the calendar. If you are weighing other SaaS tooling this quarter, our AI writing tools for SaaS roundup covers a different budget line.
Red flags: you have no legacy workload to move, your plan depends on a general availability date the vendor has not published, or you expect fixed monthly pricing when the vendor says its figures are estimates.
Common mistakes
Treating the rename as a new product purchase. Existing Azure Migrate projects continue, per the product page.
Assuming "free" covers the whole project. Azure resources, AI services, GitHub Copilot and partner tools are billed separately.
Ignoring the 180-day clock on server replication, which starts charging after the free window.
Giving one person Owner rights instead of splitting planning and execution roles. Our self-verifying agentic AI explainer covers why a human check still matters for agent output.
Frequently asked questions
Is Azure Catalyst a new product or the renamed Azure Migrate?
Azure Catalyst is formerly Azure Migrate, according to Microsoft's product page, and existing Azure Migrate projects continue to work in it.
Is Azure Catalyst free?
The core tools are free, but Azure resources, AI services, GitHub Copilot and partner tools are billed separately, and server replication costs $25 per month per instance after the first 180 days.
Does Azure Catalyst cover mainframe workloads?
No, the pages we opened do not list mainframe workloads for Azure Catalyst, while AWS Transform names a mainframe modernization agent.
Is Azure Catalyst generally available?
The vendor has not published a general availability date for Azure Catalyst as a whole, and its Learn overview labels the Azure Catalyst agent as preview.
Should a small SaaS team act this month?
Act only if you already have a migration or modernization project on the calendar; otherwise, watch the preview features and revisit when Microsoft states general availability.
Which is cheaper, Azure Catalyst or AWS Transform?
The pages do not allow a like-for-like comparison, because Azure Catalyst's core is free while AWS Transform meters its custom transformation agent, so the answer depends on your workload and how many minutes it needs.
Conclusion: decide this month or wait
For most small B2B software companies, Azure Catalyst is a watch-and-plan decision. If you run a legacy VMware, Hyper-V, SQL Server, .NET or Java estate and have a move on the calendar, the free core and the 180-day window make it worth a scoped trial. If you do not, nothing on the launch pages requires you to act this month. Before you commit, see how US Tech Automations configures this review step for the approval and budget points described above.
About the Author

Helping businesses leverage automation for operational efficiency.