The Java ecosystem has moved fast this year. From the launch of JDK 26 in March to the arrival of JDK 27 in September — plus major shifts in tooling and the Spring ecosystem — here's everything worth knowing from the past six months.

JDK 26: Performance, Startup Speed, and HTTP/3

JDK 26 landed on March 17, 2026 as a non-LTS release, following the six-month cadence established after JDK 25 (the current LTS). While enterprises on conservative upgrade cycles will largely skip this version, developers interested in performance gains have plenty to explore[citation:16].

The headline features are now locked in:

JEP 522 — G1 GC Throughput

Reduces synchronization overhead between application and GC threads. Expect 5–15% throughput gain on object-heavy workloads[citation:1][citation:18].

JEP 516 — AOT Object Caching

Project Leyden's ahead-of-time object cache now works with any GC, including ZGC. Cuts startup and warmup time[citation:1][citation:8].

JEP 517 — HTTP/3 Support

The built-in HTTP Client API now supports HTTP/3 over QUIC. Opt-in, with faster handshakes and no head-of-line blocking[citation:3][citation:18].

JEP 500 — "Final Means Final"

Deep reflection modification of final fields now issues a warning. Future releases will make this an error by default[citation:3][citation:8].

JEP 530 — Primitives in Patterns

Fourth preview of primitive types in pattern matching, instanceof, and switch. Reduces boxing overhead[citation:1][citation:8].

JEP 525 — Structured Concurrency

Sixth preview of the API that treats groups of concurrent tasks as a single unit of work[citation:8][citation:16].

Startup boost: With JEP 516, Spring Boot applications like PetClinic showed a 41% startup improvement when pre-loading 21,000 classes into the AOT cache[citation:18].

JDK 27: Post-Quantum Cryptography and G1 as Default

JDK 27 released on September 15, 2026, delivering nine JEPs — including four preview features and one incubator[citation:6]. This release continues the cadence while pushing Java further into AI-era workloads and modern cryptography.

Key additions include:

JEP 527 — PQ Hybrid Key Exchange

TLS 1.3 now supports hybrid key exchange combining classical and post-quantum algorithms. Apps using javax.net.ssl benefit without code changes[citation:6].

JEP 523 — G1 Default Everywhere

The G1 garbage collector becomes the default in all environments, including client-side. Previously server-only[citation:6].

JEP 534 — Compact Object Headers Default

Object headers shrink from 96 to 64 bits by default on 64-bit systems. Reduces heap size and improves data locality[citation:6].

JEP 532 — Primitives in Patterns (Fifth Preview)

Continued refinement of pattern matching with primitive types, now in its fifth preview[citation:6][citation:11].

Project Leyden: Startup Time Gets Serious Attention

Project Leyden continues to deliver tangible improvements to Java's cold-start problem — a long-standing criticism compared to native-image or scripting languages. JDK 26's JEP 516 extended ahead-of-time object caching to work with any garbage collector, not just G1/Serial/Parallel[citation:8][citation:18].

The GC-agnostic format means teams can train the cache with one GC and deploy with another. For containerized workloads where cold starts directly affect cost and user experience, this is a meaningful step forward[citation:16].

Tooling: IntelliJ IDEA Leads the Way

IntelliJ IDEA 2026.1 shipped in March with day-one support for JDK 26, including preview features. The release also added built-in support for multiple AI agents (Codex, Cursor, and any ACP-compatible agent), Spring runtime insight for inspecting injected beans without pausing execution, and Kotlin-aware JPA for detecting framework pitfalls[citation:10].

In a notable shift, JetBrains released Java & Kotlin by IntelliJ IDEA as a preview extension for VS Code and Cursor in August. The extension brings IntelliJ's language intelligence — code completion, navigation, analysis, and refactoring — to VS Code-based editors, with support for Maven, Gradle, and Bazel[citation:20].

Meanwhile, Eclipse IDE 2026-06 added JDK 26 support with improved debugging (statement-level step filtering, bytecode instruction highlighting) and a new formatter option for Java text blocks[citation:15]. The September release (2026-09) brought an experimental unified-diff mode and a new dialog for creating Compact Source Files, aligning with JDK 25's script-like simplicity[citation:5].

Spring Ecosystem: Boot 4.x Matures

The Spring ecosystem has been active throughout 2026. Spring Boot 4.0.x reached its sixth patch release (4.0.6) in April, while Spring Boot 3.5.x continues receiving maintenance updates through June 2026[citation:4][citation:14].

Spring Boot 4.1.0 released in June with new features including Spring gRPC support and reduced memory consumption for toByteArray() calls[citation:9]. The 4.2.0-M1 milestone arrived in August with AMQP 1.0 support and image-based build cache for Buildpacks[citation:19].

Perhaps most significantly, Spring AI 2.0 reached general availability in June 2026, marking Spring's formal entry into the AI application framework space[citation:9].

The Java Verified Portfolio

Alongside JDK 26, Oracle introduced the Java Verified Portfolio (JVP) — a curated set of tools, frameworks, libraries, and services with official Oracle support. The initial portfolio includes commercial support for JavaFX (the UI framework) and Helidon (a microservices framework)[citation:1].

Oracle also announced plans to align Helidon's release cadence with the Java release cycle and proposed bringing it into the OpenJDK project[citation:1].

What's Next

VersionStatusRelease
JDK 27ReleasedSeptember 2026
JDK 28In developmentMarch 2027
JDK 25Current LTSSeptember 2025

JDK 28 is expected in March 2027, continuing the six-month cadence. For teams that need long-term stability, JDK 25 remains the LTS choice, with updates through 2030[citation:2].

The through-line across all of this: Java is getting faster to start, more expressive with pattern matching, and more relevant for AI-era workloads — while keeping its enterprise reliability intact. The platform's commitment to backward compatibility means teams can adopt these improvements incrementally, without rewriting what already works.