The cloud engineer myth

Who this is for

• You've passed AZ-900, AWS Cloud Practitioner, or both, and the applications aren't converting to interviews.
• You're about to spend £2,000 on an associate-level cert because someone told you that's the thing standing between you and a £45–55k cloud role.
• You're a hiring manager wondering why every junior cloud applicant looks identical and none of them can actually do the job.

What the role advert says vs what the job is

The job spec for "junior cloud engineer" in 2026 reads roughly like this. AWS or Azure experience. Terraform. A bit of Kubernetes. CI/CD pipelines. Linux administration. On-call rotation. Salary £45–55k.

Read that list back to yourself. There is no version of that role that's a sensible first job in tech. Every line in it assumes you've already done the underlying work somewhere else. Terraform doesn't make sense to you until you've broken a few production deployments. Kubernetes doesn't make sense until you've operated a service that wasn't on it. On-call doesn't make sense until you've been woken up once.

The reason these adverts exist at the "junior" level isn't that employers want true beginners. It's that calling the role "junior" keeps the salary band lower while still attracting candidates with the underlying chops, usually sysadmins or infra engineers who've picked up cloud in their current job and want a title bump.

Where the myth came from

Around 2017 to 2019, AWS and Azure were genuinely new at scale. Nobody had production experience because production cloud at scale was three years old. Companies hired aggressively from sysadmin, from networking, from development, and yes, occasionally straight off the helpdesk if you had the cert. People who timed that window well had unusually good outcomes, and a content economy formed around explaining how they did it.

That content has not been retired. The bootcamps still sell the 2018 path. The YouTube ads still show the £70k salary. The cert providers still imply the cert is the thing that gets you hired. None of those parties have any incentive to update the story, because the story is what sells the product. The market quietly closed and nobody told the marketing.

Hiring managers know. Recruiters mostly know. Applicants are the last group to find out, usually after eight months of polite rejections.

The bit nobody admits

Cloud is not really a domain. It's a deployment substrate that sits on top of a domain. Cloud security is security plus cloud. Cloud networking is networking plus cloud. Cloud data engineering is data engineering plus cloud. The "+ cloud" part is six to twelve months of focused learning for someone who already has the underlying domain. It's three to five years for someone who's trying to learn both from scratch at the same time.

This is why the people who do well as cloud engineers were almost always something else first. Sysadmin first, cloud second. Networking first, cloud second. Software engineer first, cloud second. The cloud part is what they bolted on to make themselves more useful. It wasn't the starting point.

What people get wrong

First mistake. Stacking cloud certs without ever touching a real environment that costs real money. AZ-900, then AZ-104, then AZ-305, then maybe a Terraform Associate. Twelve months of evenings spent in flashcard apps and you can recite the difference between a service principal and a managed identity but you've never deployed anything that anyone other than you uses. That CV doesn't beat a sysadmin who's spent two years quietly Terraforming bits of their company's infra.

Second one. Confusing "cloud" with the role you actually want. People say "I want to do cloud" the way they used to say "I want to do computers". It isn't specific enough to be useful. Do you want to build platforms? Run them? Secure them? Optimise the bill? Migrate legacy systems off them? Those are four different jobs, hiring from four different feeder pools, with four different skill stacks. Pick one before you spend another £500 on training.

Third one. Assuming the cert opens the door. Certs filter, they don't open. A cert plus no evidence is the same outcome as no cert plus no evidence. The cert plus three small finished projects you can talk about is a different story.

The realistic route in

For someone starting in 2026 with no infrastructure background, the cloud engineering path is roughly this. Eighteen to thirty months in IT support or junior sysadmin. Cloud picked up on the side during that period, against a free tier, with three small projects finished and visible. Internal lateral or external move into a platform, infra, or junior cloud role around month thirty. The actual "cloud engineer" title shows up somewhere between year four and year six, usually after a promotion or a job change.

If that timeline feels long, that's because it is. It's also the honest version of the timeline the people in the job followed, almost without exception. The faster routes you've seen advertised are survivor bias, employer relationships that don't apply to you, or fiction.

When to ignore this

Two real exceptions. You're already a developer or sysadmin with two plus years of production experience, in which case "going cloud" is a six to twelve month pivot and the cert really does help. Or you're applying into a graduate scheme at a hyperscaler or consultancy, which has its own ladder and its own rules.

Outside those, the cloud-engineer-as-first-job story is one of the most expensive misreadings of the market currently in circulation, and the cost is paid by people who least could afford it.

Where this connects on POST

The pathways page covers the realistic feeder routes into cloud, including IT support to sysadmin to cloud engineer. The certifications section explains what each cloud cert actually signals at market level, and which ones to skip. For the hiring side of the same problem, read Wrong Pool.