From Code to Cloud: Java’s New Chapter

Welcome back to your newsletter!

In a recent InfoQ interview, Kevin Dubois of Red Hat and Thomas Vitale of Systematic outlined a fresh view of modern Java delivery. Their ideas touch on serverless performance, developer experience, security, artificial intelligence and cloud sovereignty.

Here are the key takeaways.

Java Finds Its Place in Serverless

Kevin Dubois explained that Quarkus now moves heavy work to build time and supports GraalVM native images. This change lets Java functions start fast and consume little memory, matching or beating Node.js cold starts.

Thomas Vitale noted that Spring Boot offers the same option, so teams loyal to Spring can also explore serverless without performance fears.

Developer Experience Becomes Central

Dubois traced the journey from fragile weekend releases to near instant feedback. Containers and automated pipelines now allow a laptop environment to mimic production.

Vitale highlighted Testcontainers, a library that starts real services such as PostgreSQL or Grafana whenever an application launches. The result is less configuration, more time for feature work and fewer late‑night fixes.

Security Shifts Left by Adding Capability

Vitale stressed that shifting left should mean giving developers tools, not extra burdens. Both Quarkus and Spring Boot integrate with CycloneDX, so a software bill of materials appears automatically whenever code is built.

Dubois added that modern IDEs warn about outdated or risky packages as soon as they enter a project. Teams stay compliant without hunting for information.

Blending AI with Deterministic Logic

Vitale is contributing to Spring AI, a project that wraps language models behind simple HTTP calls. Dubois is doing similar work with Quarkus and LangChain4j. Both speakers warned that large language models produce creative but unpredictable output.

Guardrails now validate every request and response, and they limit what an agent can do if it calls external tools. This balance keeps innovation in scope while protecting data and systems.

Local Models Grow Practical

Running a language model without the internet sounded impossible a year ago. Today tools such as Ollama and Podman AI Lab make it routine. Dubois and Vitale each described coding on long flights while a local model generated summaries and snippets. Local inference supports greener computing and frees developers from constant cloud access.

Open Source Underpins Cloud Sovereignty

European rules on digital sovereignty demand freedom to move workloads. Dubois argued that open source Kubernetes and OpenTelemetry already give that portability. Vitale agreed, noting that open standards make vendor lock-in optional.

In the AI world, open models from communities like Mistral or IBM Granite follow the same principle, although full transparency of training data is still a work in progress.

Key Takeaways

Dubois and Vitale show that the fundamentals of solid engineering still apply. Rapid feedback, secure supply chains and transparent tooling remain crucial even as Java embraces serverless and artificial intelligence.

The message is clear: use new frameworks thoughtfully, keep proven practices in place and always design for an enjoyable developer experience.

If these insights could shape your own architecture or team workflow, reply and let’s explore the possibilities.

Is your team hiring?

We work with global businesses across the UK, Germany and the US to help them find top talent quickly.

Reach out to us for a recruitment strategy session

🔥 Hot Jobs by develop 🔥

Data Engineer

Engineering Manager - MERN

Software Engineer

Senior DevSecOps Engineer

See More

😎 Our Social Stories 😎

Code vs Communication

The marketing team behind develop

Are tech interviews finally moving beyond whiteboards and binary trees?

Are you following us?

Click here to be the first to get our updates!