Everpure (NYSE: PSTG), a company focused on storage and data management, has introduced new solutions to advance Enterprise AI adoption. The company announced Evergreen//One for FlashBlade//EXA and revealed plans for the beta launch of Everpure Data Stream later in 2026.

These offerings aim to reduce cost and complexity, which often delay Enterprise AI projects. As a result, organizations can move faster from pilot stages to full-scale deployment.

Evergreen//One (EG1) for AI now extends to FlashBlade//EXA. It delivers high performance, scalability, and throughput required for large-scale AI training and inference. In addition, Everpure Data Stream will enable automated data pipelines, removing the need for manual data movement from ingestion to inference.

According to Kaycee Lai, Vice President of AI at Everpure, many enterprises struggle to productionize AI because they treat it as a standard workload. Therefore, the company is focusing on breaking down data silos and enabling scalable AI infrastructure.

Meanwhile, Everpure has aligned FlashBlade//EXA with modular NVIDIA STX reference architecture. This supports next-generation AI factories powered by the Vera Rubin platform. The integration includes BlueField-enabled storage controllers and context memory architectures. Consequently, it optimizes the AI pipeline from data preparation to long-context inference.

Furthermore, the architecture addresses the demands of giga-scale inference. It provides low-latency data access required for agentic workflows and multi-step reasoning systems.

Recent benchmarks from SPECstorage Solution 2020 and MLPerf confirm the platform’s performance. FlashBlade//EXA achieved the highest score in the SPEC Storage AI_Image benchmark. It also supported 6,300 simultaneous AI jobs, demonstrating strong concurrency.

In comparison, FlashBlade//EXA moves data twice as fast as competing solutions. At the same time, it occupies less than half a rack. It also sustains over 90% GPU utilization across large NVIDIA Hopper clusters, as validated by MLPerf workloads. As workloads increase, the system scales linearly, reducing idle compute time.

Additionally, NVIDIA-Certified Storage validation has been extended to FlashBlade//EXA. This creates a clear path toward NVIDIA Cloud Partner certification, aligned with NVIDIA reference architectures.

To support these capabilities, Everpure Data Stream simplifies data orchestration and curation. It ensures that high-quality, AI-ready data flows seamlessly into infrastructure. As a result, organizations can shorten the time from pilot to production.

The solution is co-engineered with Supermicro and built on the NVIDIA AI Data Platform reference design. By combining hardware and software-defined storage, enterprises can unlock greater value from their data.

Moreover, Everpure supports accelerated platforms such as NVIDIA RTX PRO 6000 Blackwell Server Edition. Support will also expand to the NVIDIA RTX PRO 4500 Blackwell Server Edition GPU.

Overall, Everpure emphasizes that success in Enterprise AI depends on continuous data optimization rather than a one-time upgrade.

Related post

View all