Mainframe Modernization Services Market

On 17 July 2025 the US Government Accountability Office reported that 11 of the most critical federal legacy systems, run by 10 agencies and aged between 23 and 60 years, still lacked complete modernization plans in eight cases, and that both Treasury systems on the list still run on COBOL and assembly code. The GAO report is the plainest public statement of a problem every large bank, insurer and government shares: the code that settles payments and pays benefits is old, and the people who wrote it are retiring. Douglas Insights values the mainframe modernization services market at USD 8.41 billion in 2025 and forecasts USD 22.7 billion by 2035, a compound growth rate of 10.43%. The receipt is about 3,420 organisations running an active modernization programme in 2025, each spending an average of USD 2.46 million a year on outside services. Programmes grow 6.8% a year as more of the roughly 10,000 enterprises that still operate mainframes start work, and spend per programme rises 3.4% a year as projects move from assessment into code conversion and data migration. The report sits within Douglas Insights coverage of enterprise software and follows the published Douglas Insights research methodology.

What counts as mainframe modernization services, and what stays outside?

Mainframe modernization services are the paid outside work that moves, rewrites, connects or re-engineers applications and data running on IBM Z and compatible mainframes, worth USD 8.41 billion in 2025 on Douglas Insights’ count. Six service lines make up the market. Re-hosting and re-platforming moves COBOL, PL/I and assembler workloads largely unchanged onto x86 servers or a public cloud using emulation and rehosting runtimes. Code refactoring and language conversion rewrites that code into Java, C# or other modern languages, increasingly with AI tools. Data migration and modernization moves VSAM, IMS and Db2 data into relational and cloud databases. Assessment and modernization consulting inventories code, maps dependencies and builds the business case. Integration and API enablement exposes mainframe transactions to mobile apps and cloud services without moving them. Testing and post-migration managed services prove the new system matches the old one and run it afterwards.

The market spans three approaches: keep and extend on the mainframe, where the platform stays and gains APIs and DevOps tooling; hybrid mainframe and cloud, where some workloads move and the core stays; and full migration off the mainframe. Mainframe hardware sales, software licences, routine outsourced operation of an unchanged mainframe and in-house staff time are excluded.

Why did the GAO’s July 2025 audit of federal legacy systems matter to mainframe modernization services?

The GAO audit mattered because only 3 of the 11 critical federal systems had fully documented modernization plans. The federal government spends more than USD 100 billion a year on IT, about 80% of it on operating and maintaining existing systems, and the GAO asked Congress to consider requiring plans by law. Douglas Insights estimates that government and public sector buyers spent USD 1.35 billion on mainframe modernization services in 2025, 16% of the market, and that agency programmes grow faster than the market once plans are mandated.

How does a mainframe modernization project turn millions of lines of COBOL into code a bank can change?

A mainframe modernization project typically runs in four stages over 18 to 48 months, and Douglas Insights estimates a large bank estate holds 20 million to 100 million lines of COBOL, PL/I and assembler across 2,000 to 10,000 programs. The first stage is discovery: automated tools parse the code, the job control language and the database calls to map which programs touch which data, often finding that 20% to 40% of the code is dead and can be retired. The second stage picks a path for each application: wrap, rehost or convert. The third stage is conversion and data migration, done in waves so the business never loses its system of record. The fourth stage is parallel running and testing, in which the old and new systems process the same transactions and outputs are compared line by line until they match. That testing stage often takes 25% to 35% of the programme budget, which is why testing and post-migration managed services form a segment in their own right.

What drives banks, insurers and agencies to fund mainframe modernization services?

Four drivers lift the number of active mainframe modernization programmes by 6.8% a year, and the retirement of COBOL skills is the largest. Douglas Insights estimates that the median COBOL developer in North America and Europe is past 55, and that about 10% of that workforce leaves each year while universities train only a small fraction of replacements. Contract rates for experienced COBOL and CICS developers reached USD 90 to USD 150 an hour in 2025 in the United States. When a bank cannot hire people to change its core code, every new product or regulatory change becomes slower and costlier, so the skills gap converts directly into modernization budgets.

Cloud strategy is the second driver. Hyperscalers now sell dedicated mainframe migration tooling and fund partner programmes, and Kyndryl reported USD 1.9 billion of hyperscaler-related revenue in its fiscal year to March 2026, up 59%, in its fiscal 2026 results. Douglas Insights estimates that about 38% of 2025 mainframe modernization services spending went to programmes with a hybrid mainframe and cloud target, and that this share rises to about 45% by 2035.

AI-assisted code conversion is the third driver. Generative tools that read COBOL and propose Java, document undocumented programs and write test cases have cut the cost of conversion per line by 30% to 50% on the pilot projects Douglas Insights tracked, which brings refactoring within reach of mid-sized banks and insurers that could not afford a manual rewrite. The same tools shorten discovery from months to weeks, so more assessments turn into funded projects: Douglas Insights estimates that about 42% of assessments started in 2025 moved into a paid build phase within a year, against about 30% in 2021.

Regulatory pressure on operational resilience is the fourth driver. The EU Digital Operational Resilience Act has applied since 17 January 2025, and supervisors in the United Kingdom, Australia, Singapore and India ask banks to prove they can recover critical services within set tolerances. Douglas Insights estimates that banking and financial services buyers spent USD 3.87 billion on mainframe modernization services in 2025, 46% of the market, and that resilience findings from supervisors triggered or accelerated about one in five new bank programmes that year.

What holds back mainframe modernization services when projects run late?

Three restraints are built into the mainframe modernization services forecast, and project failure risk is the first. Douglas Insights estimates that about one large programme in three overruns its budget by more than 30%, and several high-profile bank and state agency migrations have been paused or reversed after cutover problems. Every visible failure pushes other boards to extend on the mainframe instead, which caps spend per programme.

The mainframe’s own renewal is the second restraint. IBM’s newest machines run more AI inference and more transactions at lower cost per unit than the systems they replace, so the business case for leaving the platform weakens for high-volume core workloads. Douglas Insights removes about 0.8 points a year from programme growth to reflect organisations that choose to stay and extend, and holds keep and extend on the mainframe at about 34% of 2025 spending.

Scarce delivery capacity is the third restraint. The same retiring COBOL experts are needed to convert the code, and Douglas Insights estimates the global pool of consultants with both mainframe and cloud skills at under 60,000 people in 2025. Rationed senior staff hold spend growth to 3.4% a year rather than 5% to 6%.

Which mainframe modernization service carries the value, re-hosting or refactoring?

Re-hosting and re-platforming carries the most value at 26% of 2025 spending, USD 2.19 billion, while code refactoring and language conversion is the fastest-growing segment at 14.6% a year.

Re-hosting and re-platforming is worth USD 2.19 billion in 2025. Moving code largely unchanged onto emulation software is the fastest and least risky way to cut mainframe licence and hardware costs, so it is the first step for many insurers and public agencies.

Code refactoring and language conversion is worth USD 1.77 billion in 2025 and grows fastest at 14.6% a year, because AI-assisted conversion has lowered the cost per line enough for rewriting to beat rehosting on a ten-year cost basis.

Data migration and modernization is worth USD 1.35 billion in 2025 and grows 10.5% a year, as banks want mainframe data in cloud data platforms for analytics and AI even when the transactions stay put.

Assessment and modernization consulting is worth USD 1.18 billion in 2025. Every programme starts with an inventory and a business case, and many organisations commission several before committing.

Integration and API enablement is worth USD 1.09 billion in 2025 and grows 9.7% a year, driven by organisations that keep the mainframe and connect it to mobile banking, open banking, cloud services and, increasingly, the tokenized settlement rails sized in our Blockchain Market report.

Testing and post-migration managed services is worth USD 841 million in 2025. Parallel running and regression testing are unavoidable, and providers that run the new estate afterwards earn multi-year fees.

How do approach and end user split mainframe modernization spending?

By approach, Douglas Insights estimates keep and extend on the mainframe at USD 2.85 billion (34%) of 2025 spending, hybrid mainframe and cloud at USD 3.20 billion (38%) and full migration off the mainframe at USD 2.36 billion (28%). By end user, banking and financial services account for USD 3.87 billion (46%), insurance for USD 1.26 billion (15%), government and public sector for USD 1.35 billion (16%), and retail, travel and other users, including airlines, retailers, telecom operators and manufacturers, for USD 1.93 billion (23%).

Which region spends the most on mainframe modernization services, and how quickly is Asia Pacific rising?

North America spends the most on mainframe modernization services at 41.2% of 2025 value, USD 3.47 billion, while Asia Pacific grows fastest at 12.3% a year. North America leads because the largest US banks, insurers, federal agencies and state benefit systems hold the world’s biggest COBOL estates, and the GAO audit keeps public programmes in view. The region grows 9.6% a year to USD 8.67 billion by 2035.

Europe spends USD 2.25 billion in 2025 and grows 9.9% a year, led by banks in Germany, France, the United Kingdom, the Nordics and Spain, with DORA pushing resilience testing and exit planning for critical systems.

Asia Pacific spends USD 1.80 billion and grows 12.3% a year, the fastest rate, because Japanese banks and insurers face a sharp retirement wave among mainframe engineers, Australian and Indian banks are rebuilding core platforms, and Indian service providers deliver much of the world’s conversion work from the region. Latin America spends USD 530 million and grows 11.0% a year, led by large Brazilian and Mexican banks. The Middle East and Africa is the wildcard at USD 362 million and 10.47% a year, where Gulf banks and South African insurers and banks decide between rehosting and new cloud-native cores.

Which companies win mainframe modernization contracts, and on what strength?

Douglas Insights estimates that the five largest mainframe modernization services providers hold about 36% of 2025 value, led by IBM Consulting at about 9%.

Advantage in mainframe modernization services rests on two assets: a bench of engineers who read both COBOL and Java, and tooling that automates discovery, conversion and testing. IBM Consulting argues for keeping core workloads on IBM Z and wrapping them in APIs. Kyndryl, which runs many of the world’s mainframes, reported USD 3.5 billion of Kyndryl Consult revenue in fiscal 2026, up 18%, and launched agentic AI services for mainframe modernization that year. Accenture, TCS and Infosys win large migrations to cloud with conversion factories in India. Specialists such as Rocket Software, which acquired OpenText’s application modernization business in 2024, mLogica, Astadia and Heirloom Computing supply rehosting runtimes and conversion tools, often as subcontractors to the large firms. Hyperscaler funding often decides which integrator gets the deal.

What price does a mainframe modernization programme carry per year and per line of code?

The average mainframe modernization programme spent USD 2.46 million a year on outside services in 2025, and Douglas Insights expects about USD 3.44 million by 2035 as more programmes reach the conversion and data migration stages. Assessments cost USD 150,000 to USD 1.5 million depending on estate size. Automated code conversion is priced at about USD 1 to USD 4 per line of code, against USD 15 to USD 25 per line for a manual rewrite. Rehosting a mid-sized application portfolio typically costs USD 2 million to USD 10 million, while a full core banking migration for a large bank can exceed USD 200 million over five years. Fixed fees per line with outcome guarantees are spreading, moving overrun risk to the supplier.

Why does AI-assisted code conversion reshape the economics of mainframe modernization services?

AI-assisted code conversion cut total refactoring programme cost by 20% to 35% on the mainframe modernization projects Douglas Insights tracked in 2025. Tools from IBM, AWS, Microsoft and the large integrators explain undocumented COBOL, suggest Java equivalents and generate unit tests. The effect is two-sided. Cheaper conversion lowers revenue per line, but it pulls many more applications into scope, and Douglas Insights models the volume effect as larger: refactoring grows 14.6% a year even as price per line falls. The tools also change the keep-or-leave debate. IBM’s z17 mainframe, announced on 8 April 2025 and generally available from 18 June 2025, can run up to 450 billion AI inference operations a day, and IBM uses the same AI to help clients modernize code in place rather than move it, which supports the keep and extend on the mainframe approach.

Which laws make mainframe modernization services a board-level matter for banks?

Operational resilience law is what makes mainframe modernization services a board matter, and the most important is the EU Digital Operational Resilience Act, Regulation (EU) 2022/2554, applied from 17 January 2025. DORA requires financial entities to map critical ICT systems, test recovery, manage third-party ICT risk and keep exit strategies, which forces banks to document ageing mainframe estates and plan for their replacement or support. The UK operational resilience regime set a 31 March 2025 deadline for firms to remain within impact tolerances, and US banking supervisors apply third-party risk management guidance to outsourced mainframe work.

How large could mainframe modernization services grow by 2035 if AI conversion stalls or matures fast?

The base scenario takes the mainframe modernization services market to USD 22.7 billion by 2035, with a range of USD 16.1 billion to USD 30.7 billion. The base case combines 6.8% programme growth with 3.4% annual growth in spend per programme for 10.43% a year. The stall scenario assumes visible project failures and a stronger case for staying on new mainframes, setting the legs at 4.6% and 2.0% for USD 16.1 billion. The acceleration scenario assumes AI conversion tools mature quickly and laws in the United States follow the GAO call for mandatory plans, setting the legs at 8.8% and 4.6% for USD 30.7 billion. Each 1-point change in programme growth moves the 2035 figure by about USD 2.22 billion.

Douglas Exclusive: the mainframe exit and retention ledger

The mainframe exit and retention ledger tracks, for about 1,200 named banks, insurers, agencies and large enterprises in 24 countries, the size of each mainframe estate, the chosen approach, the lead provider, the programme stage and any announced exit date to show which estates come to market and when. Database targets for migrated data are covered in the Relational Databases Software Market report, the delivery tools teams use during conversion in the Agile Project Management Tools Software Market report, and public-sector replacement platforms in the Enterprise Permitting and Licensing Software Market report.

Methodology and receipts: how do 3,420 modernization programmes add up to USD 8.41 billion?

The mainframe modernization services model is built bottom-up from programmes. The headline receipt is 3,420 active programmes multiplied by USD 2.46 million of annual outside spend, giving USD 8.41 billion for 2025. The count starts from about 10,000 enterprises in 24 countries that operate IBM Z or compatible mainframes, then applies the share with an active, funded modernization programme by industry and country, cross-checked against provider disclosures, public tenders and GAO and agency plans. Spend per programme is set by stage and approach from contract values, tender awards and provider revenue, then split into six service lines. The forecast compounds 6.8% programme growth and 3.4% growth in spend per programme from the 2025 base to about 6,600 programmes and USD 22.7 billion in 2035.

Sources

1. US Government Accountability Office Information Technology: Agencies Need to Plan for Modernizing Critical Decades-Old Legacy Systems (GAO-25-107795) (2025)
2. IBM Newsroom IBM z17: The First Mainframe Fully Engineered for the AI Age (2025)
3. Kyndryl Holdings, Inc. Kyndryl Reports Fourth Quarter and Full-Year 2026 Results (2026)
4. EUR-Lex, Publications Office of the European Union Regulation (EU) 2022/2554 (Digital Operational Resilience Act) (2022)