Weka
Weka is an AI-native data platform company founded in 2013 by Omri Palmon, Liran Zvibel, and Maor Ben Dayan—three Israeli entrepreneurs who previously worked on XIV, a storage startup acquired by IBM for $350 million in 2007.
Profile
Software-defined, scale-out storage systems for AI model training, inference, and high-performance computing workloads.
Weka is an AI-native data platform company founded in 2013 by Omri Palmon, Liran Zvibel, and Maor Ben Dayan—three Israeli entrepreneurs who previously worked on XIV, a storage startup acquired by IBM for $350 million in 2007. The company builds parallel file systems and storage infrastructure optimized for accelerated AI and GPU workloads. In May 2024, Weka raised $140 million in Series E funding at a $1.6 billion post-money valuation, reaching unicorn status.
The round was led by Valor Equity Partners and included commitments from NVIDIA, Generation Investment Management, and others. By end of 2024, Weka crossed $100 million in annual recurring revenue and grew its customer base to 300+ enterprises, including 12 of the Fortune 50. The company's core product evolved from a storage platform called Weka into NeuralMesh, a software-defined storage architecture announced in June 2025, with a GPU-native variant called NeuralMesh Axon launched in limited release in July 2025.
Notable customers include Stability AI (which reported 93% GPU utilization and 95% cost reduction per terabyte), Cohere, Hugging Face, Together AI, and CoreWeave. In May 2025, Weka earned Gartner's Customers' Choice recognition for File and Object Storage Platforms, with a 4.9 out of 5-star rating and 98% of reviewers recommending the platform. However, Weka underwent significant organizational turbulence in early 2025: in December 2024, three senior executives departed (President Jonathan Martin, CFO Intekhab Nazeer, and CRO Jeffrey Gianetti); in February 2025, the company laid off approximately 50 employees while planning to hire 120 new staff as part of a go-to-market restructuring; in May 2025, newly appointed CFO Hoang Vuong (hired in December 2024) exited just four months into the role. CEO Liran Zvibel has publicly positioned the restructuring as strategic alignment to capitalize on what he calls a 'once-in-a-generation opportunity' in generative AI and enterprise AI markets.
Who buys this
- Enterprise AI teams training and fine-tuning large language models
- GPU-native AI/ML infrastructure providers and neocloud operators
- Fortune 50 and mid-market enterprises deploying generative AI factories
- High-performance computing and scientific research organizations
- Media and entertainment companies running visual effects rendering and video encoding workloads
Publicly disclosed clients
- Stability AI
- Cohere
- Hugging Face
- Together AI
- Contextual AI
- CoreWeave
- Nebius
Strengths and what to watch
Strengths
- Proven performance advantage in metadata IOPS and training throughput; Weka delivered ~2.75 million metadata IOPS versus DDN's ~520,000, a 5× advantage in IO500 benchmarks
- Strong customer satisfaction: 4.9/5 stars on Gartner, with 98% of reviewers recommending the platform and customers reporting up to 20x improvement in inference token generation
- Unicorn valuation and >$100M ARR with major investor backing including NVIDIA, Valor, and Generation Investment Management
Watch for
- Volatile C-suite turnover: three senior executives departed December 2024; two CFO transitions within five months (Nazeer→Vuong→Heuberger); new CRO, CPO, and CSO appointed April 2025, signaling ongoing realignment
- Recent layoffs (50 staff in February 2025) combined with aggressive rehiring (120 planned) create execution risk; company is restructuring go-to-market and R&D simultaneously amid rapid product pivots
- Market concentration in AI infrastructure; success is tightly coupled to generative AI and GPU acceleration demand, and customers are primarily hyperscalers and Fortune 50 enterprises—a narrow ICP with potential customer churn risk if AI spending normalizes
Recent moves
- 1y ago Weka launches Bengaluru Center of Excellence to expand R&D and customer success in India
- 1y ago NeuralMesh Axon moves storage to GPU servers' local SSDs for microsecond-latency AI inference and training
- 1y ago Weka introduces NeuralMesh, a dynamic mesh architecture for enterprise AI and agentic workloads with microsecond latency
- 1y ago Weka named 2025 Customers' Choice in Gartner Peer Insights for File and Object Storage Platforms with 4.9/5 rating
Key Information
- Industry
- GPU Cloud / Infra
- Founded
- 2013
Frequently Asked Questions
What is Weka?
Weka is an AI-native software-defined storage platform built for accelerated AI and GPU workloads. Founded in 2013, it provides parallel file systems and infrastructure optimized for large language model training, inference, and high-performance computing. The company reached unicorn status in 2024.
How much funding did Weka raise?
Weka raised $140 million in Series E funding in May 2024, achieving a $1.6 billion post-money valuation and unicorn status. The round was led by Valor Equity Partners and included investments from NVIDIA, Generation Investment Management, and others. By end of 2024, Weka exceeded $100 million in annual recurring revenue.
Which companies use Weka?
Notable Weka customers include Stability AI, Cohere, Hugging Face, Together AI, CoreWeave, and Contextual AI. By late 2024, Weka served 300+ enterprises, including 12 of the Fortune 50. These customers span AI infrastructure, model development, and GPU-native computing sectors seeking optimized storage for training and inference.
What are Weka's performance advantages?
Weka delivers approximately 2.75 million metadata IOPS compared to competitors' 520,000, a 5× performance advantage in IO500 benchmarks. Customers report 20× improvements in inference token generation and up to 93% GPU utilization. The platform earned 4.9/5 stars on Gartner in 2025 with 98% of reviewers recommending it.
What is NeuralMesh?
NeuralMesh is Weka's next-generation software-defined storage architecture announced in June 2025, purpose-built for enterprise AI and agentic workloads requiring microsecond latency. NeuralMesh Axon, launched in July 2025, moves storage to GPU servers' local SSDs for ultra-low-latency AI inference and training. General availability is planned for fall 2025.
Has Weka experienced leadership changes?
Weka underwent significant leadership transitions in early 2025. Three senior executives departed in December 2024: President Jonathan Martin, CFO Intekhab Nazeer, and CRO Jeffrey Gianetti. CFO Hoang Vuong exited in May 2025 after four months. CEO Liran Zvibel characterized the restructuring as strategic alignment for AI market opportunities.
How Weka compares
Direct head-to-head against 3 competitors. Picked by 7wData.
Weka
- Positioning
- Software-defined, scale-out storage systems for AI model training, inference, and high-performance computing workloads.
- Customer segments
- Enterprise AI teams training and fine-tuning large language models
- Strengths
- Proven performance advantage in metadata IOPS and training throughput; Weka delivered ~2.75 million metadata IOPS versus DDN's ~520,000, a 5× advantage in IO500 benchmarks
- Watch for
- Volatile C-suite turnover: three senior executives departed December 2024; two CFO transitions within five months (Nazeer→Vuong→Heuberger); new CRO, CPO, and CSO appointed April 2025, signaling ongoing realignment
- Recent moves
- Weka launches Bengaluru Center of Excellence to expand R&D and customer success in India
DDN
- Positioning
- Parallel file system and AI storage platform sold to HPC clusters, GPU cloud operators, and government research institutions at scale.
- Customer segments
- National labs, AI cloud providers, Fortune 500 enterprises, and government research centers running GPU-dense model training workloads.
- Strengths
- Decade-long deployments across top-tier national labs and AI cloud operators, validated at exabyte scale with Lustre-based parallel storage.
- Watch for
- HPC heritage means sales cycles and support models are optimized for research clusters, not general enterprise AI teams.
- Recent moves
- January 2025: Blackstone invested $300M at a $5B valuation to fund enterprise AI market expansion.
VAST Data
- Positioning
- Unified AI data platform combining storage, database, and compute orchestration for large-scale inference and training workloads.
- Customer segments
- AI cloud operators, large enterprise AI teams, sovereign cloud programs, and research institutions running petabyte-scale multi-modal workloads.
- Strengths
- Single universal namespace for file, object, and database access across petabyte deployments, eliminating data tiering overhead.
- Watch for
- Rapid expansion from storage into compute and database increases integration complexity. Analysts question platform cohesion at $30B valuation.
- Recent moves
- April 2026: $1B Series F closed at $30B valuation with NVIDIA participation. Azure partnership announced late 2025.
Pure Storage
- Positioning
- All-flash enterprise storage platform targeting AI factories, GPU cloud operators, and HPC workloads at petabyte scale.
- Customer segments
- Enterprise AI teams, GPU cloud operators including CoreWeave and Foundry, research labs, and sovereign cloud projects needing predictable latency.
- Strengths
- Evergreen subscription model delivers non-disruptive hardware upgrades. DirectFlash architecture reduces write amplification on NVMe media.
- Watch for
- All-flash-only approach carries higher CapEx per terabyte vs. software-defined tiered alternatives. NAND price volatility compresses customer budgets.
- Recent moves
- March 2025: Launched FlashBlade//EXA targeting over 10TB/s read throughput for AI and HPC in a single namespace.
Sources
- www.weka.io — Company overview, products (NeuralMesh, NeuralMesh Axon), notable customers (Stability AI, Cohere, Hugging Face, Together AI, Contextual AI, CoreWeave, Nebius)
- techcrunch.com — Series E $140M funding, $1.6B valuation, unicorn status, CEO quote on market opportunity, >300 customers, $100M+ ARR, ~400 employees
- www.prnewswire.com — Series E details: $140M raised May 15 2024, Valor Equity Partners lead, existing investors (NVIDIA, Generation Investment Management, Atreides, etc.), Antonio Gracias joins board
- blocksandfiles.com — February 2025 ~50 employee layoff, 120 new hires planned, go-to-market restructuring, CEO quote on GenAI opportunity
- blocksandfiles.com — CFO Hoang Vuong resignation May 10 2025, appointed December 2024, timeline of leadership departures (Nazeer, Martin, Gianetti in December 2024)
- blocksandfiles.com — Executive leadership changes April 2025: Brian Froehling as CRO, Ajay Singh as CPO, Nilesh Patel as CSO
- www.weka.io — Current leadership team: Liran Zvibel (CEO), Maor Ben-Dayan (Chief Architect), Amit Pandey (Executive Chairman), Meini Heuberger (CFO)
- www.prnewswire.com — Gartner 2025 Customers' Choice recognition, 4.9/5 star rating, 98% recommendation rate, works with 12 Fortune 50 companies
- www.weka.io — CEO strategic letter: unicorn milestone, >$100M ARR, more than doubled revenue third consecutive year, focus on enterprise AI and GPU acceleration
- www.weka.io — NeuralMesh announcement June 2025, purpose-built for agentic AI, microsecond latency, general availability fall 2025