# See What's Possible with WEKA

Explore WEKA customer case studies. See how enterprises like Krea AI, Luma AI, and Firmus eliminate bottlenecks and accelerate GPU workloads.

## Proven Results Across Every Workload

- **90%** Reduced storage costs — Center for AI Safety cut cloud storage spend after switching to WEKA.
- **34%** Lower Time to First Token — Firmus sped up responses on multi-turn coding-agent workloads.
- **35%** Faster Model Training — Stability AI shortened training times by removing storage bottlenecks.

## Stories

- [Rendering the Future](/customer/krea-ai) — Krea — **~40%**
  Krea AI sustains roughly 40% Model FLOPs Utilization
- [Building a More Reliable Foundation for Research](/customer/advanced-research-computing) — Advanced Research Computing
  Oxford ARC gains serviceability, fault isolation, and one platform across InfiniBand and Ethernet
- [WEKA and Andromeda Partner to Power AI Workloads at Global Scale](/news/weka-and-andromeda-partner-to-power-ai-workloads-at-global-scale) — Andromeda AI
  Andromeda delivers consistent storage performance across 50+ compute providers worldwide
- [Engineering Out the Ceiling](/customer/firmus) — Firmus — **6.5x**
  Firmus breaks the DRAM ceiling, serving up to 6.5x more tokens on every watt
- [Singtel Digital Infra Co’s RE:AI and WEKA partner to deliver sovereign AI infrastructure across ASEAN](/news/singtel-reai-and-weka-partner-to-deliver-sovereign-ai-infrastructure) — Singtel : REAI
  Singtel RE:AI brings sovereign AI infrastructure to governments and enterprises across ASEAN
- [Powering Tomorrow's AI Innovations Today](/customer/ytl-ai-cloud) — YTL AI Cloud
  YTL AI Cloud powers Malaysia’s first locally developed large language model
- [Powering Transformative AI Research with WEKA](/customer/deakin-university) — Deakin University — **10x**
  Deakin University trains 10x faster, speeding earlier cerebral palsy detection in infants
- [Accelerating AI Adoption at Scale](/customer/neysa-ai) — Neysa
  Neysa eliminates storage bottlenecks and delivers enterprise-grade reliability
- [Accelerating Every Experiment with WEKA](/customer/luma-ai) — Luma AI — **40x**
  Luma AI starts training runs 40x faster, keeping H100s creating instead of waiting
- [Powering Cultural AI Independence](/customer/siam-ai) — Siam AI
  Siam AI delivers breakthrough performance for Thailand’s indigenous AI cloud services
- [DCAI Expands AI Infrastructure Offering with WEKA’s Integrated Storage Services](/news/dcai-expands-ai-infrastructure-offering-with-wekas-integrated-storage-services) — DCAI — **140PB**
  DCAI gives Denmark’s Gefion supercomputer 140 PB of sovereign storage at sub-millisecond latency
- [Streamlining Training of Foundational AI Models](/customer/lg-ai-research) — LG AI Research
  LG AI Research scales its multi-node AI models more effectively
- [Reaching the Most Ambitious AI Transformation Goals with WEKA® NeuralMesh™](/customer/core42) — Core42
  Core42 builds its AI accelerator portfolio on high-performance NeuralMesh storage
- [Powering the Next-Generation AI Cloud with WEKA NeuralMesh](/customer/nebius) — Nebius
  Nebius scales its AI cloud from petabytes to exabytes with microsecond latency
- [Powering the Future of Audio-Visual Intelligence](/customer/dolby-laboratories) — Dolby Laboratories
  Dolby moves data seamlessly between on-premises and AWS to maximize GPU utilization
- [Eliminating Storage Bottlenecks at Enterprise Scale](/customer/tensorwave) — Tensorwave
  Tensorwave maximizes GPU utilization and speeds time-to-insight across its AI cloud
- [Powering Open-Source AI at Unmatched Speed and Scale](/customer/together-ai) — Together AI
  Together AI cuts latency for its inference engine, serving 500,000+ AI developers
- [Cohere Eliminates Storage and GPU Bottlenecks at Scale with NeuralMesh](/customer/cohere) — Cohere — **10x**
  Cohere checkpoints 10x faster across its AI training and inference clusters
- [Scaling Cutting-Edge Speech Synthesis and Generative Audio Models](/customer/elevenlabs) — ElevenLabs
  ElevenLabs scales its voice AI models with fast, simple access to data
- [Empowering Sustainable AI Innovation with High-Performance Data Solutions](/customer/iren) — IREN
  IREN delivers lightning-fast data access for GPU cloud services on 100% renewable energy
- [Powering Digital Transformation with NeuralMesh](/customer/novartis) — Novartis
  Novartis accelerates cloud-based research while cutting HPC costs
- [Igniting a Robotics Revolution](/customer/physical-intelligence) — Physical Intelligence — **80%**
  Physical Intelligence cuts data infrastructure costs 80% and speeds robotics training 10-15%
- [Democratizing AI for a Collaborative Future](/customer/hugging-face) — Hugging Face
  Hugging Face accelerates AI model loading into GPUs on AWS with low latency
- [Streamlining Its Color Science and Finishing Workflows](/customer/picture-shop) — Picture Shop
  Picture Shop gets reliable playback with zero caching, freeing artists to create
- [Dead & Company: Dead Forever at Sphere](/customer/dead-and-company) — Dead & Company
  Dead & Company powers its most ambitious and immersive concert experience ever
- [Flexibility in the Cloud Elevates Innovation and Research Impact](/customer/the-wharton-school) — The Wharton School
  Wharton scales its elastic HPC cloud so researchers meet publication deadlines
- [Changing How the World Works Through AI](/customer/contextual-ai) — Contextual AI — **38%**
  Contextual AI triples performance on key AI use cases while cutting cloud storage costs 38%
- [Nightly Sky Surveys Generate Petabytes of Astronomical Data](/customer/vera-c-rubin-observatory) — Vera C. Rubin Observatory
  Vera Rubin Observatory transforms our view of the universe with a 10-year southern sky survey
- [Leading AI Innovation Beyond Digital Limits](/customer/upstage) — Upstage
  Upstage removes GPU-to-storage bottlenecks to launch high-performance LLMs faster
- [Accelerating Large Language Model Training in the Cloud](/customer/stability-ai) — Stability AI — **35%**
  Stability AI cuts storage costs 95% per terabyte while training models 35% faster
- [Accelerating AI Safety Research at Lower Cost in the Cloud](/customer/center-for-ai-safety-cais) — Center for AI Safety
  CAIS accelerates storage performance 5x, supporting 200+ researchers on one GPU cluster
- [Pushing Live Entertainment to Impossible Heights](/customer/u2-at-sphere) — U2:UV
  U2 moves 200-300 GB of video every minute to power its immersive Sphere shows
- [Pioneering Sustainable AI Infrastructure](/customer/yotta) — Yotta
  Yotta builds India’s most powerful and affordable AI cloud on sustainable infrastructure
- [Serving Continuous Data to the Fastest GPUs on the Planet](/customer/applied-digital) — Applied Digital
  Applied Digital boosts GPU efficiency and affordability for its generative AI cloud
- [Post-Production Profitability Is All About the Cloud](/customer/parliament-vfx) — Parliament VFX — **95%**
  Parliament VFX keeps 95% of its data in an economical tier without losing performance
- [Genomics England Fuels Critical Genome Research](/customer/genomics-england) — Genomics England
  Genomics England fuels its 5 million genome mission with 41 PB in a single namespace
- [Cerence Builds a Cost-Effective, High Throughput Bridge to The Cloud](/customer/cerence) — Cerence
  Cerence gains superior cost-efficiency, scalability, and flexibility with WEKA
- [Simplifying and Accelerating Scientific Research](/customer/omrf) — OMRF — **10x**
  OMRF runs research jobs 10x faster, cutting some from 70 days to seven
- [Accelerating Biomedical Research & Discovery](/customer/swiss-institute-bioinformatics) — SIB Swiss Institute of Bioinformatics
  SIB encrypts sensitive biomedical data in flight with no performance impact
- [Powering a Global VFX Studio in the Cloud on NeuralMesh](/customer/preymaker) — Preymaker
  Preymaker connects artists across three continents in one global namespace on AWS
- [Untold Studios Chooses Weka on Amazon Web Services for Simplicity in Cloud Implementation](/news/untold-studios-chooses-wekaio-matrix-amazon-web-services-simplicity-cloud-implementation) — Untold Studios
  Untold Studios scales hybrid cloud storage across on-prem and AWS without the overhead