Estimate project
← All posts

Did We Get It Right? Revisiting Our "IT Trends 2021" Predictions Five Years Later

We released “IT Trends to Watch in 2021” in December 2020, at a time of enormous global uncertainty. Now, five years on in 2025, it’s a good moment to revisit our predictions and check which ones hit the mark, which were too optimistic, and where things turned out a bit differently.

3 min read
Machine LearningArtificial IntelligenceTech TrendsData StreamingCybersecurity

In December 2020, when we released “IT Trends to Watch in 2021”, the world was facing enormous uncertainty. COVID-19 had pushed everyone into remote work. Digital transformation had suddenly become a pressing matter, and the IT landscape seemed to be changing shape before our eyes.

Now, five years on in 2025, it’s a good moment to revisit our predictions and check which ones hit the mark, which were too optimistic, and where things turned out a bit differently.

What we got right

AI and ML in every corner

We expected AI and machine learning to leave the research labs and make their way into real-world products, even reaching the edge, mobile devices, IoT, and embedded systems.

That prediction was right on target.

Whether it’s image recognition running directly on smartphones, on-device language models, smart cameras, or AI-powered apps, edge AI is no longer a thing of the future. Companies are more and more often combining cloud-based models with lightweight inference at the edge. We may have somewhat underestimated how quickly AI would grow and how much generative AI would accomplish, but we called the overall direction correctly.

Streaming data and real-time in everything

In 2020, we wrote about the move from batch processing to streaming data pipelines. Kafka and Flink were already picking up momentum back then.

Jump ahead to today: real-time analytics, event-driven architectures, and streaming platforms are standard across fintech, e-commerce, logistics, and IoT. The tooling has matured enormously, and “real-time” is now something people expect by default rather than a luxury.

That prediction has held up nicely.

Cybersecurity as a shared responsibility

We forecast that cybersecurity would become a top priority, particularly in software supply-chain security and DevSecOps practices. Unfortunately, the industry confirmed this sooner than we would have wished.

The SolarWinds attack, Log4Shell, and many breaches involving open-source dependencies drove the point home: security isn’t a checklist but part of day-to-day engineering. Now that security is shifting left, hardly any serious team ships code without building security into CI/CD and dependency scanning.

This prediction wasn’t merely accurate – it turned into a necessity.

Remote work and a digital-first culture

We argued that remote work was not a stopgap for the pandemic but a lasting change.

Today, hybrid and remote work are the rule rather than the exception. The tools have got better, distributed culture has matured, and “digital-first” is taken for granted in every industry. The one caveat is that the world is moving toward a hybrid model more than we originally anticipated, with big companies publicly announcing return-to-office policies.

Even so, the broad prediction stands: remote work isn’t going away, at least for part of the week.

Where we were (partly) off

Blockchain past the hype

Back in 2020, we were hopeful that blockchain would mature and move beyond the crypto hype toward enterprise adoption and decentralized identity.

Five years later, that mainstream enterprise breakthrough still hasn’t arrived.

There are certainly interesting applications in supply chain tracking, finance, and digital identity, but blockchain never became the foundation of everyday enterprise systems. It remained a niche instead – valuable in particular contexts, yet not the all-purpose disruptor we had pictured.

Verdict: a promising technology that is still searching for the right problems to tackle.

IoT and 5G

We wrote that IoT was about to take off, with 5G as the key enabler. That turned out to be… optimistic.

IoT has grown steadily, and industrial uses (particularly in manufacturing and logistics) are doing very well. On the consumer side, however, fragmentation remains a nightmare: too many standards and too many “smart” devices that can’t communicate with one another.

5G, too, proved to be more of an evolution than a revolution – excellent for bandwidth and latency, but not the game-changing platform we had in mind at the time. The ecosystem is maturing, only more slowly than we expected.

Cloud-agnostic dreams versus cloud reality

In 2020, we pictured a world in which hybrid and cloud-agnostic setups would liberate teams from vendor lock-in. In theory it’s achievable; in practice, much less so.

Kubernetes, Terraform, and IaC have made moving across clouds easier, but the pull of cloud-specific services (such as BigQuery, Lambda, or DynamoDB) is too strong. Most companies are multi-cloud on paper, yet locked in through their design decisions.

The lesson: cloud-agnostic is still an ideal rather than the norm.

Main takeaways & reflections five years on

Looking back, I’m proud of how many things we got right, particularly regarding AI, data streaming, security, and culture.

What we underestimated wasn’t wrong as such; it simply moved more slowly, turned out more complicated, or was driven by economics and regulation rather than by technology alone. These areas may still mature in the coming years.

If these five years have taught us one thing, it’s that forecasting technology is difficult, and getting the timing right is harder still.