WEKA

What WEKA does

WEKA is a privately held data platform company founded in 2013 and headquartered in Campbell, California, that builds NeuralMesh, a software-defined, parallel file system with an adaptive mesh architecture purpose-built for AI workloads. The platform delivers microsecond-latency storage at petabyte-to-exabyte scale across on-premises, public cloud, and hybrid environments and is protected by more than 170 patents. Around the NeuralMesh core the company layers NeuralMesh Axon (GPU-native AI storage), Augmented Memory Grid (which reportedly extends effective GPU memory 1000x by streaming KV cache from NVMe flash to GPU HBM over NVIDIA GPUDirect Storage), NeuralMesh AIDP (an AI Data Platform built on the NVIDIA AI Data Platform reference design with UI/API for ingestion, embedding, tokenization, and vector indexing), and the WEKApod family of turnkey hardware appliances (Prime and Nitro). The product targets the 'AI memory wall' — the gap between exploding model context windows and limited GPU HBM — and has disclosed validated performance including 4.2x more tokens per GPU on CoreWeave, 90%+ GPU utilization for Stability AI training, and 7.5M read / 1.0M write IOPS on an eight-node OCI H100 cluster.

WEKA's customer base spans AI cloud providers and neoclouds (CoreWeave, Nebius, Singtel RE:AI, Firmus, Neysa, ResetData), frontier AI labs (Cohere, Stability AI, Hugging Face, Eleven Labs, Together AI, Contextual AI), national labs and HPC centers (SLAC, DCAI, UKAEA, RIKEN), and large enterprises, with the company stating that 30% of the Fortune 50 are customers. Revenue comes from five streams: subscription licensing of NeuralMesh software, hardware sales of WEKApod appliances, usage-based consumption on the Oracle Cloud Marketplace (OCI is the exclusive cloud launch partner), channel/OEM resale (with Supermicro, ASUS, Dell, and historically HPE), and professional services. Distribution combines direct enterprise field sales, OEM/embedded bundles, regional channel partners such as Glocomp in Malaysia, and validated reference architectures for NVIDIA DGX BasePOD, DGX SuperPOD, and NVIDIA HGX H100/H200/B200 systems.

Cumulatively WEKA has raised approximately $348 million across six disclosed rounds since 2014, including a $73M round in January 2022 led by Hitachi Ventures, a $135M Series D in November 2022 led by Generation Investment Management, and a $140M Series E in May 2024 led by Valor Equity Partners that took post-money valuation to $1.6 billion. CFO Intekhab Nazeer has publicly targeted $500 million in annual recurring revenue. The company has 251–500 employees and remains venture-backed with no disclosed IPO plans.

Name

WEKA

Legal name

WekaIO, Inc.

Website

https://weka.io

Company type

Private

Founded year

2013

Operating status

Operating

Headcount range

251–500 employees

Short description

WEKA provides a software-defined AI data platform (NeuralMesh) that extends GPU memory 1000x to address the AI memory wall. The platform serves AI cloud providers, foundation model labs, and research institutions through subscriptions, turnkey appliances, and cloud marketplace distribution.

Ownership category

akta.pro rank

Product category

AI Data Platform Software

NAICS

Computer Systems Design and Related Services (54151), Computer Systems Design and Related Services (5415)

SIC

Services-Computer Integrated Systems Design (7373), Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

AI Memory & Storage for Training (HBM/DDR/SSD/Object Storage) (HDAAAAAF)

akta.pro secondary industries

On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB), AI Compute Cloud & GPU-as-a-Service (HDAAAAAK), Decentralized GPU/AI Compute Networks (training/inference, GPU renting, model hosting) (FSAPALAC), DePIN Orchestration, Scheduling & Middleware (resource discovery, job scheduling, workload orchestration) (FSAPALAK)

Keywords

• AI data platform
• Parallel file system
• GPU memory extension
• AI storage infrastructure
• High-performance storage

Headquarters

HQ city

Campbell

HQ country

United States

HQ region

North America

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Marketing or Sales, Supply Chain, Infrastructure, Operations

Revenue model

1. Software Licensing (NeuralMesh Subscription): Recurring software license / subscription revenue from NeuralMesh data platform software sold to enterprise and AI cloud provider customers. Includes NeuralMesh, NeuralMesh Axon, Augmented Memory Grid, NeuralMesh AIDP, and Multitenancy modules.
2. Hardware Appliance Sales (WEKApod): One-time / capital hardware revenue from WEKApod, WEKApod Prime, and WEKApod Nitro appliances — turnkey AI storage appliances purpose-built for AI and HPC workloads, sold alongside or integrated with the NeuralMesh software.
3. Cloud Marketplace / Usage-Based Consumption: Consumption-based revenue through Oracle Cloud Marketplace (and other cloud platforms) where customers consume NeuralMesh and Augmented Memory Grid on a usage / capacity basis. OCI is WEKA's exclusive cloud launch partner.
4. Channel / OEM Resale Revenue: Revenue share from OEMs and system integrators (Supermicro, ASUS, Dell, HPE historically, Glocomp) that resell or co-deliver WEKA-powered AI infrastructure bundles and AI Data Platform solutions.
5. Professional Services & Deployment Support: Services revenue tied to enterprise AI factory deployment, including architectural validation, integration with NVIDIA reference designs, and managed deployment of NeuralMesh AIDP appliance-style systems.

Pricing tiers

Go-to-market motion6 records

Distribution channels6 records

Marketing channels8 records

Core offering

WEKA develops and sells NeuralMesh, a cloud-native, software-defined data platform that delivers high-throughput, low-latency file, object, and block storage purpose-built for AI and high-performance workloads. The platform is augmented by add-on modules (NeuralMesh Axon, NeuralMesh AIDP), the Augmented Memory Grid capability that extends GPU memory via token warehousing, and delivered as software, as turnkey WEKApod appliances, or as consumption on the Oracle Cloud Marketplace.

Product overview

WEKA offers a single unified AI data platform built around its core NeuralMesh software platform, delivered as a platform-plus-modules architecture. The core platform is NeuralMesh, a software-defined adaptive mesh data platform that provides high-throughput, low-latency file, object, and block storage with native GPUDirect Storage, NIXL, and NVIDIA AI Enterprise integration. On top of this core, WEKA layers add-on modules: NeuralMesh Axon for automated AI data pipelines and NeuralMesh AIDP (AI Data Platform) for unified, API-driven ingestion, embedding, tokenization, and vector indexing. These modules are complemented by the Augmented Memory Grid feature, which extends effective GPU memory for LLM inference via token warehousing and KV cache offload. The platform is delivered either as software or as turnkey hardware appliances in the WEKApod family, including the new WEKApod Prime and WEKApod Nitro systems, with multitenancy available as a built-in platform feature for shared AI/ML environments.

Differentiator

Problem solved

Functional benefit

Brands

• NeuralMesh: WEKA's flagship software-defined storage platform purpose-built for AI workloads.
• NeuralMesh Axon
• WEKApod
• Augmented Memory Grid
• NeuralMesh AIDP

Products and services

• NeuralMesh Software-defined, adaptive mesh data platform for AI and high-performance workloads, providing unified high-throughput, low-latency storage across file, object, and block protocols. Foundation layer for all WEKA products.
• NeuralMesh Axon GPU-native AI storage that embeds NeuralMesh directly on GPU servers, delivering 20x improvement in time-to-first-token and 90%+ GPU utilization for training workloads. Early adopters include Cohere and CoreWeave.
• NeuralMesh AIDP (AI Data Platform) Integrated AI data platform built on the NVIDIA AI Data Platform reference design that operationalizes enterprise AI factories. Delivered as a pre-validated appliance-style system that compresses AI factory deployment from months to minutes, with 6.5x more tokens per GPU for inference workloads.
• WEKApod Turnkey hardware appliance family delivering the NeuralMesh software platform in a pre-integrated, rack-scale system for enterprise and AI factory deployments.
• WEKApod Prime New high-end WEKApod appliance offering 65% better price-performance via intelligent data placement, designed to scale AI/HPC workloads and improve GPU utilization while addressing datacenter resource constraints.
• WEKApod Nitro Performance-optimized WEKApod appliance that doubles performance density for demanding AI inference and training workloads, with high-throughput GPU-direct storage.
• Augmented Memory Grid Software-defined memory extension technology that streams KV cache directly between NVMe flash and GPU HBM over RDMA using NVIDIA GPUDirect Storage, turning storage into a petabyte-scale token warehouse for AI inference. Extends GPU memory by 1000x, reduces time-to-first-token by up to 20x, and sustains 96-99% KV cache hit rates.
• Multitenancy for NeuralMesh Multi-tenant architecture on NeuralMesh with secure, policy-driven data isolation and per-tenant performance guarantees for AI cloud providers serving multiple customers on shared infrastructure.

Quantifiable outcome

• 1000x GPU memory extension versus DRAM-only configurations (validated on OCI)
• +11 more outcomes

Named customers17 records

Segments9 records

Ideal customer profiles5 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration20 records

AI capability9 records

Feature7 records

Partnerships

Twelve partnerships are on record, tiered flagship, core and minor.

• Singtel RE:AIflagshipStrategic or Co-development Partner · 11 June 2026MOU signed in June 2026 to build sovereign AI infrastructure across Singapore and ASEAN. Combines Singtel's AI cloud, Paragon network orchestration platform, and NVIDIA Centre of Excellence with WEKA's NeuralMesh data infrastructure, integrated into Singtel's GPU-as-a-Service offering.
• Dell TechnologiescoreStrategic or Co-development Partner · 29 May 2026Collaboration on the £45M UK Sunrise AI supercomputer for fusion energy research (6.76 ExaFLOPS, 1.4MW, Culham Campus), alongside AMD, Intel, University of Cambridge, and UKAEA. WEKA is also a storage partner in Dell's PowerEdge-based AI factory offerings.
• Glocomp SystemscoreChannel Partner/ Reseller/ Distributor · 6 May 2026Exclusive end-to-end AI infrastructure partner for WEKA and NVIDIA in Malaysia. The 'Malaysia AI Starter Pack' combines WEKA's NeuralMesh with NVIDIA accelerated computing, compressing enterprise AI deployment timelines from months to ~2 weeks, delivered via Glocomp's 150+ resellers and system integrators.
• ASUScoreOEM/ Whitelabel/ Licensing Partner · 17 March 2026NVIDIA-Certified storage partner for ASUS AI POD and XA VR721-E3 rack-scale systems on NVIDIA Vera Rubin NVL72. Co-marketed as part of ASUS's fully liquid-cooled AI infrastructure announced at Computex 2026 and GTC 2026.
• Spectro CloudcoreTechnology or Integration · 17 March 2026Joint integration of PaletteAI with NeuralMesh to simplify and accelerate deployment of NVIDIA's AI Data Platform across data centers and edge environments. Kubernetes orchestration layer for NeuralMesh AIDP.
• SupermicrocoreOEM/ Whitelabel/ Licensing Partner · 16 March 2026OEM/embedded partner for seven AI Data Platform solutions built on NVIDIA reference architectures. WEKA is one of seven storage ecosystem partners (alongside Cloudian, DDN, Everpure, IBM, Nutanix, VAST Data). Validated configurations for NeuralMesh AIDP. Supermicro, DDN, and NVIDIA also plan a mobile AI Factory demonstration.
• Red HatcoreStrategic or Co-development Partner · 16 March 2026Roadmap to extend enterprise integration with Red Hat OpenShift for consistent lifecycle management and governance across on-prem and cloud as part of NeuralMesh AIDP. Co-delivered with Red Hat AI Enterprise.
• Firmus TechnologiescoreTechnology or Integration · 16 March 2026AI cloud provider deploying WEKA on NVIDIA BlueField-4 STX. Reported 6.5x higher tokens/sec and 4x faster time-to-first-token at scale; jointly conducted proof-of-concept with WEKA (also reported by SiliconANGLE April 2026).
• NVIDIAflagshipStrategic or Co-development Partner · 5 January 2026Strategic technology and co-development partner of nearly a decade. WEKA is one of the first storage partners building next-generation AI storage platforms with NVIDIA BlueField-4 and the NVIDIA Inference Context Memory Storage Platform (ICMSP/STX). NeuralMesh AIDP is built on the NVIDIA AI Data Platform reference design. WEKA is also an NVIDIA investor (participated in Series E).
• Hewlett Packard Enterprise (HPE)minorChannel Partner/ Reseller/ Distributor · 7 November 2025HPE announced in late 2025 that it will stop reselling Qumulo, Scality, and WEKA storage software (sales quoting ended Jan 31, 2026; final orders by March 3, 2026) as HPE shifts to its own IP including Alletra Storage MP and the OEM'd VAST Data filesystem.
• Danish Centre for AI Innovation (DCAI) — GefioncoreTechnology or Integration · 23 October 2025Launched one of Europe's first fully sovereign, end-to-end encrypted AI infrastructure platforms in Copenhagen, Denmark, integrating 140 PB of storage (80 PB NVMe on NeuralMesh) with the Gefion AI supercomputer. Plans to leverage Augmented Memory Grid for LLM inference workloads.
• Oracle Cloud Infrastructure (OCI)flagshipStrategic or Co-development Partner · 18 September 2025OCI is WEKA's exclusive cloud launch partner. Joint benchmarks validate NeuralMesh Axon and Augmented Memory Grid on OCI bare-metal H100 clusters, achieving 10x more concurrent users, 10x higher token throughput, 7x more tokens per GPU, and 20x faster TTFT. WEKA is also available on the Oracle Cloud Marketplace.

Scale indicators14 records

Recent moves6 records

Expansion highlights8 records

Direct peers

• VAST Data: Direct competitor in AI-native parallel file systems / data platforms, offering a scale-out storage architecture targeting AI training and inference workloads. Closely comparable to WEKA's NeuralMesh in both technology approach and customer base (AI clouds, foundation model labs, HPC), and notably selected as the OEM'd replacement for WEKA inside HPE's portfolio.
• DDN (DataDirect Networks): Direct competitor in high-performance AI storage, with deep penetration in HPC and large AI factory deployments. Competes with WEKA in AI cloud and enterprise AI factory deals, including the EXAScaler parallel file system product line targeting similar NVIDIA reference architectures.

Broad incumbents

• Pure Storage: Large incumbent enterprise storage vendor extending its FlashArray and FlashBlade portfolios into AI workloads. Competes with WEKA in enterprise AI factory deals and on Oracle Cloud Marketplace, but operates a much broader storage portfolio beyond AI.
• NetApp: Incumbent enterprise storage vendor with ONTAP and a growing AI data platform strategy (including the Instaclustr and Spot by NetApp acquisitions). Overlaps with WEKA in enterprise AI deployments and financial services HPC, and previously employed WEKA's current Chief AI Officer Val Bercovici as CTO.
• Dell Technologies: Incumbent infrastructure vendor offering PowerScale as a parallel file system and AI factory reference architectures. Both a partner (UK Sunrise, PowerEdge-based AI factory bundles) and a competitor to WEKA in enterprise AI infrastructure deals.
• Hewlett Packard Enterprise (HPE): Incumbent storage and HPC vendor whose Alletra Storage MP B10000 and the OEM'd VAST Data filesystem are positioned to replace third-party parallel file systems. HPE was a historical WEKA reseller and equity investor but announced it will stop reselling WEKA in 2025–2026.
• IBM Storage: Incumbent storage vendor with Spectrum Scale (GPFS) and AI-ready infrastructure offerings that compete with WEKA in enterprise and HPC AI factory deployments, particularly in regulated industries and sovereign AI programs.
• Hitachi Vantara: Incumbent enterprise storage and infrastructure vendor (backed by Hitachi, WEKA's Series C lead investor). Competes in enterprise AI and HPC storage with its HCSF and VSP platforms, while sharing an investor relationship with WEKA.

Emerging players

• MinIO: Open-source, software-defined object storage company increasingly positioned for AI workloads (AIStor). Competes with NeuralMesh on object storage layers of AI pipelines, especially in cost-sensitive and Kubernetes-native deployments.
• Scality: Software-defined object and file storage vendor, identified alongside WEKA and VAST Data as a potential tier-1 storage software supplier for RIKEN's FugakuNEXT supercomputer. Overlaps with WEKA in HPC, sovereign AI, and large-scale research deployments.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/wekaio
• YouTubeyoutube.com/channel/UC6TDf_aLab6ykflcwEmkDLA
• Facebookfacebook.com/lifeatweka
• LinkedInlinkedin.com/company/weka-io
• Instagraminstagram.com/lifeatweka

Revenue estimate

Valuation estimate

Number of profiles

Profiles14 records

Funding overview

Funding rounds6 records

Investors25 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about WEKA