Software development is being turned completely on its head. Programmers are increasingly becoming architects who guide systems rather than write code. And this is only the beginning.
Not so long ago, a “Gunfight at the AI Corral” erupted between the AI company Anthropic and the American “Department of War.” The Pentagon demanded full control over the AI software it was using, but Anthropic refused to allow its technology to be deployed for domestic surveillance or autonomous weapons. The dispute escalated quickly. Secretary of Defense Pete Hegseth described Anthropic’s position as a “masterclass in arrogance.” The company, however, stood firm, prompting the Pentagon to remove all Anthropic software. Meanwhile, OpenAI was more than ready to step in and take over the contract.
A lot of drama, but above all it signals that the software world has fundamentally changed. In the past, when the Pentagon purchased a PowerPoint license, it could use it however it wished. Microsoft didn’t care whether it was for drafting a cafeteria menu or outlining military plans against Iran. With AI, things are entirely different. AI introduces new rules, new mechanisms, and new sensitivities. More importantly, the entire IT landscape is now shifting.
Since November, the tech world has been collectively fixated on the “METR graph,” which shows how much work a human programmer can delegate to AI without needing to step in. For years, this meant small tasks, something that might take a few minutes or up to half an hour. With Claude Opus 4.5 from Anthropic, that boundary suddenly leapt to more than seven hours of autonomous programming work. Newer versions even push that limit to fifteen hours.
In other words, AI can already handle the equivalent of nearly two full working days of programming without any human intervention.
This changes everything. Software development is being turned completely on its head. Programmers are increasingly becoming architects who guide systems rather than write code. And this is only the beginning.
For years, companies built their IT environments around stable software components lik ERP systems, CRM systems or accounting tools: solid foundations that organizations often relied on for ten or even twenty years. But with the rise of AI agents, systems capable of executing tasks autonomously, that once-stable IT landscape has turned into quicksand in just a matter of months.
Being responsible for IT has never been more exciting, but also never more challenging.
We’ve seen a moment like this before. In the early 2000s, many IT managers were used to purchasing and managing “traditional” technology. Then the internet took off, and suddenly everything revolved around websites, e-commerce, digital customer experiences, and, later, social media. The result was an explosion of “shadow IT.” Marketing departments often understood these new tools better than IT did. New roles, such as chief digital officer, emerged to bridge that gap. What followed were years of cleanup to regain control over fragmented systems.
Today, we risk a similar scenario, but on a much larger scale. Digital transformation was merely the appetizer; the main course is the immense impact of AI. Employees are rapidly adopting AI tools, experimenting with AI agents, and automating processes, often without any oversight from IT. This creates significant risks around cybersecurity, data breaches, and compliance.
This situation can quickly spiral into chaos unless companies invest in a new generation of technology leaders. IT directors who truly understand AI remain rare, and the same is true for suppliers. On one side, there’s an army of traditional IT consultants who have spent years building ERP systems but have little experience with AI. On the other, there are young AI startups that excel at models and agents but barely know how to spell ERP.
This is clearly not a smooth transition but a sharp divide. In this revolution, the focus is shifting away from technology itself and toward processes, work, and skills. The IT leaders who grasp this and who radically reinvent their organizations around it, will become more powerful than technology leaders have ever been.
This is a translation of a Dutch piece that was originally featured in Belgian business newspaper De Tijd.