AIBRIEF

NVIDIA Opens cuFile APIs, Cites Up to 3.21x Vera CPU Storage Throughput Ahead of FMS

N
NathanTechnology Editor · Technical Lead
Published · Updated
According to NVIDIA's blog published ahead of the FMS conference (Aug. 4-6, Santa Clara, California), NVIDIA announced it is open-sourcing its cuFile APIs and underlying storage software stack, naming Google, Intel, NVIDIA and Meta as inaugural maintainers. NVIDIA also cited internal benchmarks showing its Vera CPU delivers up to 3.21x higher throughput than an x86 CPU in a two-stage compression-and-encryption pipeline, alongside a 40-plus-vendor Storage-Next coalition and a DDN integration announcement.

What hardware breakthroughs does the NVIDIA Vera CPU deliver?

According to a technical blog cited by NVIDIA, benchmarks show that the NVIDIA Vera CPU — part of NVIDIA Vera BlueField-4 STX — delivers up to 3.21x higher throughput than an x86 CPU in a two-stage compression-and-encryption pipeline (E1). NVIDIA describes Vera BlueField-4 STX as "a modular, rack-scale foundation powered by the NVIDIA Vera Rubin platform, NVIDIA Vera BlueField-4 storage processors and NVIDIA Spectrum-X Ethernet networking" (E6). In other words, the throughput figure is tied to a specific three-part architecture — compute (Vera Rubin), storage processing (BlueField-4) and networking (Spectrum-X) — rather than a standalone chip claim.

Which companies are backing the cuFile open-source project?

At the FMS conference, NVIDIA announced it is open sourcing its cuFile application programming interfaces (APIs) — described as "the vertical storage software stack underneath them" — which let GPUs, not just CPUs, read from and write to storage directly (E2). NVIDIA named Google, Intel, NVIDIA and Meta as the inaugural maintainers of the newly opened project, stating the site is "open to contributions" and can be "optimized for use across various software and hardware platforms" (E3). The four-company maintainer list spans NVIDIA's own hardware ecosystem plus two of the largest hyperscale cloud operators (Google, Meta) and a rival chipmaker (Intel).

What do Storage-Next and the FMS conference schedule signal?

NVIDIA said its Storage-Next initiative includes over 40 leading storage and flash vendors — "including DDN, KIOXIA and Micron — each contributing to the next generation of AI storage technologies with NVIDIA" (E4). The announcements were timed around FMS (Future of Memory and Storage), which NVIDIA said runs Aug. 4-6 in Santa Clara, California, where the company is hosting sessions on its AI storage work (E7). The 40-vendor scale of Storage-Next places the cuFile open-sourcing and Vera CPU benchmarks within a broader coalition announcement rather than a single-product release.

How is DDN integrating storage systems in practice?

NVIDIA's blog reports that DDN is integrating SCADA with its Infinia platform (E5), a move NVIDIA frames as part of the same effort it describes elsewhere: "Storage-Next and SCADA extend NVIDIA's longstanding work on AI storage infrastructure, including on NVIDIA Vera BlueField-4 STX" (E6). DDN's chief technology officer, Sven Oehme, said in the announcement: "AI success will be defined not by how much infrastructure organizations own, but by how productively they use it. Our collaboration with NVIDIA is helping create a more direct, efficient connection between GPUs and data — keeping accelerated computing resources productive, speeding time to insight and enabling customers to achieve stronger business and financial returns from their AI investments" (E5). That statement is DDN's own characterization of the partnership's value, as quoted in NVIDIA's blog, rather than an independently verified figure.

Numbers at a glance

MetricValueEvidence
Vera CPU throughput vs. x86 (compression + encryption pipeline)up to 3.21x higherE1
cuFile inaugural maintainers4 companies (Google, Intel, NVIDIA, Meta)E3
Storage-Next vendor countover 40 (incl. DDN, KIOXIA, Micron)E4
FMS 2026 conference datesAug. 4-6, Santa Clara, CaliforniaE7

What this means

Taken together, the pieces NVIDIA disclosed in the same blog post point to a coordinated announcement rather than isolated updates: the cuFile open-sourcing (E2, E3) sits inside the 40-plus-vendor Storage-Next coalition (E4), and NVIDIA explicitly ties DDN's SCADA-Infinia integration (E5) to the same Vera BlueField-4 STX foundation (E6) that underpins the 3.21x throughput benchmark (E1). All of it was timed to the Aug. 4-6 FMS conference (E7), meaning the hardware benchmark, the API open-sourcing, the vendor coalition and the DDN integration were presented as one connected storage push rather than four separate stories.

📊 Evidence

FAQ

When and where is the FMS 2026 conference NVIDIA referenced?

Aug. 4-6, 2026, in Santa Clara, California, according to NVIDIA's blog.

Who are the inaugural maintainers of the newly open-sourced cuFile project?

Google, Intel, NVIDIA and Meta, as named in NVIDIA's blog announcement.

How many vendors are part of NVIDIA's Storage-Next initiative?

Over 40 leading storage and flash vendors, including DDN, KIOXIA and Micron, according to NVIDIA.

📎 Sources

  1. blogs.nvidia.com
N
NathanTechnology Editor · Technical Lead

Related

BRIEF

At SIGGRAPH 2026, NVIDIA Announces Cosmos 3 Edge, DGX Station and MotionBricks for Agentic and Physical AI

According to NVIDIA's SIGGRAPH 2026 blog post, the company announced the 4-billion-parameter Cosmos 3 Edge model, DGX Station systems built on the GB300 Grace Blackwell Ultra Desktop Superchip with up to 20 petaflops of FP4 compute, a Synthetic Video Detector microservice reaching 92% accuracy on uncompressed video, and MCP-based AI agent integrations across five creative-software platforms.

Nathan ·
BRIEF

Intel's New Mexico Fab Pushes Packaging Reticle Limit to 8x Today, Targets 12x+ by 2028

According to Intel's newsroom announcement, Intel's Fab 9 advanced packaging site in Rio Rancho, New Mexico has scaled its 'reticle limit' to 8 times the current industry standard, with a target of over 12 times by 2028. Intel says the facility grew from 25 employees in 1980 to 2,700 employees and 500 suppliers today, using Foveros and EMIB/EMIB-T technologies to package AI chips.

Nathan ·