WEKA

@WekaIO

NeuralMesh™ by WEKA® - The world's only storage system purpose-built for AI. Accelerate performance, deploy anywhere, grow stronger with scale.

California, USA

Joined November 2014

2KFollowing

2KFollowers

Pinned

WEKA@WekaIO · Jul 8

🚨 Live from #RAISESummit: WEKA unveils NeuralMesh Axon—breakthrough storage for exascale #AI. ⚡ 10x faster checkpointing ⚡ 20x faster time-to-first-token 📈 90%+ GPU utilization Built for LLMs, agentic AI & real-time inference. weka.ly/4lGLfzq

458

Pinned

WEKA@WekaIO · Jul 18

At #AIInfraSummit, WEKA CTO @shimonbd and @nvidia's Nave Algarici shared how we’re helping orgs scale #AI inferencing—combining WEKA’s real-world infrastructure with NVIDIA acceleration. 🎥 Watch the full convo: weka.ly/44QQGF1

Pinned

WEKA@WekaIO · Jul 17

GenAI at scale needs storage that can keep up. Our NCP-certified architectures for GB200 NVL72 & HGX systems deliver 1+ GB/s per GPU, scale to 18K+ GPUs, and run on fast, efficient Micron 9550 SSDs. Get the full breakdown. 👇 weka.ly/3IsGoUi

WekaIO's tweet card. Learn how WEKA’s NCP-certified storage for GB200 and HGX systems delivers 1 GB/s per GPU, enabling scalable, ultra-fast AI factories built for the next era.

2.0K

Pinned

WEKA@WekaIO · Jul 9

.@liranzvibel delivered a dynamic keynote session during @RaiseSummit exploring how the rise of the token economy is reshaping #AI. From scalable infra to sustainable innovation, the future of AI depends on balancing speed and efficiency. #RAISESummit

WekaIO's tweet image. .@liranzvibel delivered a dynamic keynote session during @RaiseSummit exploring how the rise of the token economy is reshaping #AI.

From scalable infra to sustainable innovation, the future of AI depends on balancing speed and efficiency.

#RAISESummit

109

WEKA@WekaIO · 5 h

.@togethercompute serves 500K+ devs with one of the fastest inference engines in #AI. To scale fast and slash latency, they leverage WEKA. 👉 See how: weka.ly/4o1tf4R

WEKA@WekaIO · 11 h

Speed is critical in #AI infrastructure—but efficiency is everything. NeuralMesh goes beyond traditional dedupe with similarity-aware optimization that makes your NVMe storage work smarter, not harder. Here's how. 👇 weka.ly/4kS6fm5

WekaIO's tweet card. NeuralMesh uses similarity-aware deduplication and intelligent compression to reduce storage needs by up to 8x—boosting AI performance without wasting NVMe.

WEKA@WekaIO · Jul 23

In GPU-heavy #AI workloads, every microsecond matters. Our Augmented Memory Grid + our new NIXL plugin for @nvidia Dynamo = blazing-fast inference at near-memory speeds. And we’re just getting started. Learn more. 👇 weka.ly/413LQDd

WekaIO's tweet card. Explore how NVIDIA Dynamo, NIXL, and WEKA accelerate AI inference, slash TTFT, and scale token warehouses to petabytes.

WEKA@WekaIO · Jul 22

AI: “I need low latency, infinite scale, and zero drama.” NeuralMesh: “Say less.” ⚡ Microsecond latency 📦 Exabyte scale 🤖 K8s-native magic Legacy storage could never. Explore the white paper. ⬇️ weka.ly/455mTc0 weka.ly/450ap5s

WekaIO's tweet card. The WekaFS addresses IT storage problems by providing a resilient distributed parallel file system that delivers performance, simplicity and scalability.

WEKA@WekaIO · Jul 22

.@nvidia GB200 systems are built to push the limits of #AI + ML. NeuralMesh—validated for NVIDIA’s NCP RA—scales to 18K+ GPUs with speed, simplicity, and power. Take a closer look at the full reference architecture. 👇 weka.ly/45gt12f

WekaIO's tweet card. Explore WEKA’s NeuralMesh™ high-performance storage for NVIDIA Cloud Partner solutions, with a reference architecture designed for the NVIDIA GB200 NVL72 system — scalable, AI-optimized storage...

WEKA@WekaIO · Jul 18

.@huggingface is making #AI accessible to all—and we're proud to help. 🤝 Running on WEKA + @awscloud, they’re accelerating model loading, boosting training & inference, and cutting latency. Learn more about how we're fueling the next wave of AI innovation. 👇…

WEKA@WekaIO · Jul 16

LLMs feeling slow and clunky? Our Augmented Memory Grid delivers up to 75X faster time to first token, slashing latency and handling huge prompts with ease—no extra GPUs required. Learn more. ⬇️ weka.ly/46O3NJH

WekaIO's tweet card. WEKA sets new inference time to first token TTFT industry benchmarks with its open source GPUDirect Storage integration for TensorRT-LLM.

WEKA@WekaIO · Jul 16

NeuralMesh is the only storage system built specifically for #AI. Want to see how it helps teams overcome performance and scalability hurdles? Take a closer look 👇 weka.ly/4lZkjLw

WekaIO's tweet card. Introducing NeuralMesh™️ by WEKA®️. The world’s only storage system purpose-built for AI. What it is, why it matters, and how it helps AI leaders overcome the biggest barriers to scale and speed.

WEKA@WekaIO · Jul 15

Boxed in by outdated infrastructure? That's a dead end for your #AI workloads. NeuralMesh is built for the scale, speed, and complexity modern AI demands. Discover how it works: weka.ly/4nL7WUR

106

WEKA@WekaIO · Jul 15

What sets #AI leaders apart? ⚡️ They move fast 📈 They scale smart 🛠️ They’ve left legacy infrastructure behind Exactly what we heard at #RAISESummit—and what we see every day with the teams building on WEKA. weka.ly/4kLAKKh

WEKA@WekaIO · Jul 14

Traditional storage slows down at scale. NeuralMesh gets faster. 100-node cluster: ~1 min recovery 50-node cluster: ~10 min Not magic—just math. 👇 weka.ly/40fZKSx

WekaIO's tweet card. NeuralMesh by WEKA rebuilds data faster as clusters grow, delivering unmatched resilience and ultra-fast recovery without performance trade-offs. Learn how it works.

WEKA@WekaIO · Jul 11

💬 “Partner solutions like WEKA’s NeuralMesh Axon… provide a critical foundation for accelerated inferencing…” — Marc Hamilton, @nvidia We're proud to power the infrastructure behind the world’s most advanced AI factories. Discover the full capabilities of NeuralMesh Axon:…

WEKA@WekaIO · Jul 11

At @RaiseSummit, we asked what sets the front-runners apart from those trailing behind in the #AI race. The answers were clear—and they align with what we see every day working with the world’s most innovative AI organizations. #RAISESummit

WEKA@WekaIO · Jul 10

The @LinusTech team pushed modern infra to the limit—calculating π to 300 trillion digits. With help from WEKA, they hit 150 GB/s throughput, ran nonstop for months, and recovered from power cuts like pros. Here’s how it worked: weka.ly/3IDBeVh

WekaIO's tweet image. The @LinusTech team pushed modern infra to the limit—calculating π to 300 trillion digits.

With help from WEKA, they hit 150 GB/s throughput, ran nonstop for months, and recovered from power cuts like pros.

Here’s how it worked: weka.ly/3IDBeVh

269

WEKA@WekaIO · Jul 10

That’s a wrap on #RAISESummit 2025! We unveiled NeuralMesh Axon, dove deep with partners, and explored how to scale enterprise #AI without compromise. 💥 Let’s continue to build the future of AI—together.

121

WEKA@WekaIO · Jul 9

What makes NeuralMesh Axon a game-changer for large-scale #AI? @liranzvibel breaks down how GPU-native storage boosts performance, slashes infra costs & maximizes GPU utilization. 🔓 Unlock the full potential of your GPUs: weka.ly/45WPRNe #RAISESummit

110

WEKA@WekaIO · Jul 9

Don’t let legacy storage slow you down. NeuralMesh Axon brings ultra-low latency, distributed storage inside GPU servers—eliminating I/O bottlenecks, boosting utilization, and cutting costs. 📄 Check out the solution brief: weka.ly/3Iz8kWi

WekaIO's tweet image. Don’t let legacy storage slow you down.

NeuralMesh Axon brings ultra-low latency, distributed storage inside GPU servers—eliminating I/O bottlenecks, boosting utilization, and cutting costs.

📄 Check out the solution brief: weka.ly/3Iz8kWi