AI integration paths, API behaviors and alignment of the Java ecosystem are among the improvements being introduced today in Java 27.
Donald Smith, VP of Product Management in the Java Platform Group, told SD Times that at Oracle, “we focus on putting innovation at the tip, so the latest versions of the JDK is where we focus on new, innovative, exciting things. Where we’re deep in the internals of the platform, we’re adopting new hardware architectures, new trends in the industry, adding new features in that direction, but the tail is where the enterprises really benefit from using Java because we focus very much on conservatism, making sure that we’re not making breaking changes at the tip when we have innovative features and implementations that have proven themselves to be stable, performant, and secure.”
What’s being done for developers
Integrating AI into enterprise applications involves coordinating models, data sources, streaming, transactions, and concurrency while calling external AI services, Smith explained. For developers integrating AI through APIs, the path is to use service SDKs and HTTP clients. Examples include vendor SDKs such as the OCI SDK that map responses to Java objects and simplify remote calls. For those running inference in‑process, the focus is on APIs and runtime improvements that reduce memory pressure and increase throughput.
Second, running AI workloads on Java will involve inferencing and, in some cases, model training. The improvements in this area target runtime efficiency, such as data movement, memory use, throughput, and latency, all of which are supported by ongoing projects and APIs – notably the Vector API, which remains in incubator while awaiting Project Valhalla. which is in JDK28 early access and is moving into the JDK mainline.
“Project Valhalla gives developers early visibility into efforts to bridge the long-standing gap between Java objects and primitives, starting with value classes and objects that could benefit data-heavy workloads in analytics, financial services, scientific computing, and AI,” the company wrote in its announcement. “Delivered in stages, the project aims to make Java data representation denser and more efficient while preserving the Java programming model. Valhalla gives developers and enterprises early visibility into future improvements for data-heavy applications, including workloads in analytics, financial services, scientific computing, and AI-enabled systems.”
Security changes surface directly in application code paths via TLS. Java 27 delivers post‑quantum hybrid key exchange for TLS 1.3, following IETF standards. Where peers support compatible algorithms, applications can negotiate hybrid key exchange and maintain forward security against anticipated quantum threats. Oracle plans backports to earlier JDK lines once stability and performance criteria are met, but Java 27 enables adoption now.
Developers will also see continuous tweaks that don’t rise to JEP status but affect daily work, such as API behaviors, garbage collection refinements, and configuration conveniences. The cumulative view since the move to a six‑month cadence shows steady integration of features and fixes across releases, not just the headline JEPs.
Aligning the Java ecosystem
Ecosystem alignment is part of the developer experience. JavaFX returns to a supported cadence to serve teams building desktop interfaces, including analytics‑oriented apps.
Longer‑horizon work remains relevant to future application performance. Project Valhalla aims to narrow the gap between primitive and value types to improve memory density and performance in data‑heavy code.
Project Leyden continues improving startup, warm‑up, and footprint—important for microservices and any deployment sensitive to cold‑start behavior.
Project Detroit leverages the Foreign Function & Memory API (Project Panama) to connect Java with CPython and V8 runtimes, handling memory and invocation details to streamline mixed‑language workflows common in AI‑adjacent systems.