Skip to content
DT Nova

Software development trends and challenges for 2026

Software development trends and challenges in 2026 are shaped less by new frameworks than by three forces: AI tools that write a growing share of code, EU rules that turn security and accessibility into legal duties, and a lasting shortage of people to build and maintain systems. This guide sums up what the public data says about each, with dates and sources, and what it means if you are buying or commissioning software rather than writing it. For founders building a software business from scratch, see how to start a software company.

Trend: AI-assisted coding is the default

Stack Overflow's 2025 survey found that 84% of respondents were using or planning to use AI tools in their development process, up from 76% in 2024, and that 51% of professional respondents used AI tools daily. Coding assistants, chat tools and agents that open their own pull requests are now part of most teams' work, and the cost of writing a first version of a feature has dropped.

The same survey shows the limits. Positive sentiment toward AI tools fell from over 70% in 2023 and 2024 to 60% in 2025. More respondents actively distrusted the accuracy of AI output (46%) than trusted it (33%), and only 3% reported highly trusting it. The most common frustration, named by 66%, was AI answers that are almost right, but not quite. Most respondents did not plan to use AI for high-responsibility tasks such as deployment and monitoring (76%) or project planning (69%), and 72% said they were not vibe coding (Stack Overflow 2025 survey, AI section).

Google's DORA research reaches a similar conclusion from the organisation's side: its 2025 report describes AI's primary role as an amplifier, magnifying an organisation's existing strengths and weaknesses, with the greatest returns coming from the underlying organisational system rather than the tools themselves (DORA 2025, State of AI-assisted Software Development).

Challenge: trusting what the tools produce

When code is quick to write, review and testing become the bottleneck. A team that generates more code than it can read ends up with more defects, more security holes and a codebase nobody fully understands. The practical answers are not new, but they matter more: small changes, mandatory review by a person who understands the domain, automated tests that run on every change, and clear rules about which data may be sent to which AI service. For buyers, the useful question to a supplier is not whether they use AI, but how they check its output. See software testing for what a testing setup covers.

Our AI in software development guide sets out working rules for using these tools safely.

Trend: regulation becomes a product requirement

For anyone selling software or connected products in the EU, 2025 to 2027 is a period of deadlines.

RuleWhat appliesFrom
European Accessibility ActAccessibility for covered consumer services, such as e-commerce, banking and passenger transport apps28 June 2025
AI Act, Article 50(1)People must be told they are interacting with an AI system, unless obvious2 August 2026
Cyber Resilience Act, reportingEarly warning of actively exploited vulnerabilities and severe incidents within 24 hours11 September 2026
Data Act, cloud switchingNo switching charges for moving to another data processing service12 January 2027
Cyber Resilience Act, in fullEssential cybersecurity requirements, including a software bill of materials11 December 2027

Sources: Directive (EU) 2019/882, Regulation (EU) 2024/1689, Regulation (EU) 2024/2847 and Regulation (EU) 2023/2854.

The Cyber Resilience Act changes the most for software makers. Its essential requirements include identifying and documenting the components in a product, including by drawing up a software bill of materials in a commonly used, machine-readable format covering at least the top-level dependencies. That turns dependency management from good practice into something a market surveillance authority can ask to see. The Act covers products with digital elements placed on the market, so custom software built only for one company's internal use is in a different position from an app or device sold to many customers; the boundary is worth checking with the supplier early.

Challenge: security of the supply chain

A typical application today is mostly other people's code: open source libraries, cloud services, AI models and the build tools that put them together. Each is a way in for an attacker and a source of urgent updates. The practical response is a maintained list of components, automated alerts for known vulnerabilities, a written process for applying security updates within a set time, and a clear owner for each system after launch. For an independent look at where you stand, see our security audit.

Trend: integration over new systems

Few companies start from nothing. Most new software work connects systems that already exist: an ERP to an online store, a scheduling app to payroll, a customer portal to the CRM. The trend shows in what gets built: APIs, data pipelines, internal tools and automations rather than large standalone systems. It also shows in what fails. When a project goes wrong, the cause is more often an unclear interface, missing data or an owner who left than the choice of programming language. See API integration for how we approach it.

Challenge: the shortage of specialists

Eurostat reports that 20.05% of EU enterprises employed ICT specialists in 2024. Of the 9.55% of enterprises that recruited or tried to recruit ICT specialists during 2023, 57.5% had difficulties filling those vacancies (Eurostat, ICT specialists: hard-to-fill vacancies). For a small or mid-sized company, hiring a full team for one system is rarely realistic. The common answers are to buy a project from an outside team, keep a small internal owner who knows the business, and agree maintenance in writing so the system does not depend on one person.

Trend: cloud cost and portability

After a decade of moving to the cloud, many companies now look harder at what they pay and how locked in they are. The Data Act requires providers of data processing services to remove obstacles to switching and, from 12 January 2027, not to charge switching fees. That makes portability a fair question to ask any supplier: can the system and its data be moved to another provider, and has anyone tried? Hosting in an EU region and owning the cloud account in your own name make the answer easier. See cloud migration consulting.

Challenge: older systems that still run the business

Much of the work in software is keeping older systems alive: a desktop application that only one person can change, an in-house ERP on an unsupported database, a website on a plugin nobody updates. These systems rarely fail all at once. They get slower to change, harder to secure and more expensive to connect. Replacing them in one step is risky; the safer route is usually to wrap them with an interface, move one function at a time and retire the old part when the new one has proven itself.

What this means if you are buying software

  • Ask how AI output is checked. Review rules, tests on every change and limits on what data goes to AI services.
  • Put security in the contract. Security updates within a set time, a component list or software bill of materials, and a named owner after launch.
  • Check accessibility duties. If the European Accessibility Act covers your service, build to WCAG from the first design, not as a fix later.
  • Own the code and the data. The repository, the cloud account and the database in your name, with a tested export.
  • Budget for maintenance. Software that is not maintained becomes a risk within a year or two; plan for it from the start.
  • Start small. A first version on real data, measured against the way you work now, tells you more than any specification.

For how projects are organised, see our guide to software development methodologies, and for budgets, custom software development cost. For a build, see software development, and for a second opinion before you commit, software consulting. For the stack itself and how long each part is supported, see software development technologies.

The short version

What is the biggest trend in software development in 2026?

AI-assisted coding. In Stack Overflow's 2025 survey, 84% of respondents were using or planning to use AI tools in their development process, up from 76% the year before, and 51% of professional respondents used them daily. The open question is no longer adoption but how to review, test and trust what the tools produce.

What are the main challenges in software development now?

Trust in AI-generated code, new EU rules that turn security and accessibility into legal requirements, the shortage of ICT specialists, and the cost of keeping older systems running and connected. Most projects that fail today fail on scope, integration and maintenance, not on the choice of framework.

How does the Cyber Resilience Act affect software projects?

If you place a product with digital elements on the EU market, the Act's reporting obligations for actively exploited vulnerabilities and severe incidents apply from 11 September 2026, with an early warning due within 24 hours of becoming aware. The rest of the Regulation, including the essential cybersecurity requirements and a software bill of materials, applies from 11 December 2027.

Is it harder to hire software specialists in Europe?

Yes, for most companies that try. Eurostat reports that 57.5% of EU enterprises that recruited or tried to recruit ICT specialists during 2023 had difficulties filling those vacancies. Many companies respond by buying projects or maintenance from an outside team instead of hiring for every skill.

What should a company buying software do differently in 2026?

Ask how AI-generated code is reviewed and tested, make security updates and a software bill of materials part of the contract, require accessibility where the European Accessibility Act applies, insist on owning the code and data with a tested export, and budget for maintenance from the start.

Sources