Six moments that defined the tech landscape in 2025

If 2024 was the year we talked about AI, then 2025 replied: “Fine. Let’s see how serious you are.” Let’s take a subjective tour of the year's most interesting tech moments.

AI started to make decisions

For a long time, AI in software development meant rather small things: suggesting the next line of code, fixing syntax, etc. By 2025, that quietly changed. We’ve seen tools like Cursor and GitHub Copilot evolve into "Agentic IDEs."

Today’s AI tools can analyse repositories, follow dependencies, and propose changes that span multiple services. A great example is the Agent Mode released in GitHub Copilot Workspace (mid-2025), which allows a developer to assign a GitHub Issue to the AI; it then drafts the plan, writes the code across multiple files, and opens the Pull Request itself (under close human review, of course!).

That shift has changed the developer’s job. In many teams, engineers write less code and spend more time reviewing what AI produces. There’s a clear advantage: work moves faster. But there’s also a side effect – the question “will AI replace developers?” is being asked more often. And more importantly, how will junior developers learn the “craft" if the AI does all the entry-level work? We should not be surprised if 2026 brings the answer.

AI moved into .NET 10

If 2024 was about "adding" AI to your project, with .NET 10, Microsoft leaned harder into making AI a first-class concern across the ecosystem (especially around identity, security, and performance).

What else changed?

• No more passwords: .NET 10 officially made Passkeys (FaceID/Fingerprint) a citizen in its identity system, and passkeys became the standard for many apps built this year.
• Future-proof security: It introduced native support for post-quantum cryptography (ML-DSA and ML-KEM algorithms), keeping your data private even against the "code-breaking" computers of the next decade.
• Eco-friendly code: .NET 10 is the most performant .NET release yet, and its runtime improvements mean software can do the same work with fewer CPU cycles. For app users, it translates into less energy, longer battery life for devices, and lower cloud power usage.

The internet proved it’s held together by three companies and a spell

We like to think of the internet as a vast, indestructible web, but 2025 reminded us it’s actually more like a single high-voltage cable that a few big players control and everyone is trying to use at once. We saw a domino effect:

• An automated update to Google's API management triggered a global crash that disrupted services like Spotify, Discord, Meet, and Maps.
• When the AWS US-East-1 region (Northern Virginia) collapsed, it dragged half the internet down with it. A DNS error prevented applications from finding the DynamoDB database, reportedly affecting hundreds of major services worldwide.
• Then there was Cloudflare, whose sequence of outages broke access to everything from X and ChatGPT to Canva and Shopify.

We’ve centralised the decentralised web, and now, when one giant trips, the rest of the world hits the floor.

Your data got a passport

For years, the "Cloud" meant your data lived everywhere and nowhere. In 2025, that changed with the rise of Sovereign AI. Countries like France (with Mistral AI) and Germany started pulling their data back behind physical borders to ensure "technological autonomy." We realised that if you don’t own the servers, you don’t really own your business. Software architecture is now less about "where is the cheapest server?" and more about "whose laws apply to this code?" It’s an expensive shift (goodbye, cheap one-size-fits-all cloud) and might end the “wild west” era of global data. In 2025, data got a passport and a border checkpoint.

The energy hit home

In 2025, we finally had to talk about the electricity bill. Between AI data centres and the 5G expansion, tech’s hunger for power became impossible to ignore. Major tech firms (including Microsoft and Google) made headlines this year by striking deals for nuclear energy (Small Modular Reactors) to power their AI grids. This was the year "Green Software" became a requirement. We’re seeing developers rewarded for writing code that uses less CPU, and that is not just to save money, but because the local power grid literally can't handle the load.

The reality check

This year, we stopped trusting everything we saw on a screen. With AI-generated videos and audio (deepfakes) becoming indistinguishable from reality, 2025 saw the rise of content provenance efforts. Major social platforms rolled out some form of AI content labelling by default. For example, Meta (Facebook/Instagram) and TikTok announced automatic labels for AI-generated posts, marking images, videos, or audio as “made with AI” whenever their systems detect AI manipulation or users flag it.

In short, 2025 gave everyone a crash course in media literacy: if you see a shocking video clip or hear a suspicious audio snippet, assume it might be AI until proven otherwise. Ironically, software is helping us figure out what’s actually human.

The shift from "More" to "Better"

If I had to summarise 2025 in one sentence, it would be this: We stopped being impressed by what tech could do and started worrying about what it is doing. In 2026, I suspect the most innovative companies won't be the ones adding more features, but the ones making our digital lives feel stable and human again.