AI coding tools have dominated conversations about the future of software development, promising faster delivery, smaller teams and huge productivity gains. But as businesses scale their use of these tools, many are realising the economics are far more complicated than they first expected.
Rising usage costs, unpredictable billing models and growing pressure on development budgets are forcing organisations to rethink whether generating more code with AI is actually the most cost-effective way to build software.
As pricing increasingly scales with developer output, the question of whether AI coding tools are delivering enough value to justify their cost is becoming harder to ignore. That pressure is pushing more businesses towards low-code platforms instead – tools designed to reduce the need for custom code in the first place, while avoiding the spiralling consumption costs that can come with it.
So, let’s look at why businesses are embracing low-code.
The AI coding reality check
For a time, the idea of using AI to write code faster than any human could – while cutting costs – seems like an obvious win for businesses focused on efficiency. But that narrative is starting to change. Uber’s CTO recently revealed that his 5,000 engineers depleted their entire 2026 token budget in just four months, while 71% of companies exceeded their AI budgets in 2025.
This is partly because AI costs are highly variable and are starting to scale with output.
Anthropic recently more than doubled its estimate for how much the average business will end up spending on its Claude Code tool, and GitHub Copilot has moved to usage-based billing, which will quickly compound across engineering teams. For SMEs, where budgets are tighter and cost predictability matters more, even modest overruns can have a real impact.
As a result, more organisations are questioning whether AI-generated code is delivering the level of return on investment they originally expected.
Why businesses are embracing low-code
At the same time, the cost of traditional software development continues to rise. Recruiting developers is expensive, onboarding and training takes months, and working with external agencies often means costs increase with every new feature request or change in scope.
Low-code platforms offer a way to reduce that pressure from multiple fronts. Teams on the business side can create and adapt applications themselves, without relying on expensive development hours. Changes that might otherwise spend weeks sitting in a development backlog can be handled directly by the people using the software every day instead.
This also allows smaller businesses to develop more advanced tools without significantly increasing their IT spend. This will be especially significant for SMEs because the kind of custom tools that were only for large IT teams are now something a small team can build and own themselves, without the unpredictable spend associated with AI coding tools.
Reducing the cost of software development
Low-code platforms are also evolving to include AI capabilities directly within the development environment, allowing organisations to build AI applications without needing machine learning expertise. For example, chatbots, automated data analysis, sales forecasting, image recognition, and customer inquiry routing can be added through visual tools.
AI models can now be integrated into existing systems with minimal custom development, meaning features that would have previously been a dedicated specialised project can now be shipped as part of a normal development cycle.
As a result, the barrier for building an application stops being about whether someone has programming knowledge or not. Employees outside of traditional development teams are able to build and maintain tools that fit their own workflows, while developers stay in control and can extend those tools with code where it’s needed.
In a market where specialised technical talent remains costly, this approach also reduces the reliance on a small pool of expensive specialists and distributes the work evenly across the organisation.
What low-code looks like in practice
One of the more understated advantages of low-code is the impact it has on how developers spend their time. By automating repetitive work like boilerplate code, setting up database schemas and building standard UI components, developers can focus more of their time on the complex, creative problems that drew them to the profession in the first place.
This also means business requirements get handled faster, less time is lost in planning cycles, and more time is spent on building. Working better with non-technical colleagues also helps make briefs clearer and leads to fewer reworks.
For example, a dashboard that gives sales teams real-time stock and revenue data can be built and shipped in days rather than weeks, allowing developers to focus on the harder problems while the rest of the business gets what it needs much quicker.
What’s more, in the coming years, fewer young people will be entering the workforce than those leaving it. For businesses already feeling the pressure of a shrinking developer talent pool, low-code platforms offer a practical way to remain competitive and profitable by building and shipping tools without depending on resources they simply don’t have.
Organisations already adopting low-code platforms are not trading quality or capability for speed. Instead, many are shipping faster, working better across teams, and saving their expertise for the work that delivers the most value.
Businesses that successfully combine low-code platforms with traditional software development will be more likely to build and adapt faster than those still treating every application as a fully bespoke project.
