๐ Big things have been happening at #SC26.
AI is changing what data centers need, and WEKApod is changing whatโs possible in every rack.
๐ Here's what more looks like: https://t.co/KwOwj5n5E5
Up to 60% of GPU compute can go to redoing work.
WEKAโs Alan McSeveney joined @VentureBeatโs Rob Strechay to explore how KV cache efficiency can change that.
Catch up on the full convo ๐ https://t.co/ppxK3nerCI
๐ญ GPU utilization isnโt the full picture.
The Inference Economics Scorecard measures what matters: useful work, efficiency, and real results. Nine metrics. One standard. https://t.co/EC4Egp3iyW
@ScaleoutSean Zoom e-mails saying someone had joined a scheduled zoom started dribbling in hours after each meeting ended,the first couple had me worried I was ignoring a call.
๐ Welcome to WEKA, Nic Morales, our first Chief Customer Officer!
With experience at @cohere, @alteryx, and @IBM, Nic is a great addition to the team. https://t.co/zh6LTpHC9S
Incredible numbers proven with WEKAPod3. Proof that our software defined architecture can fully exploit faster and denser hardware as it comes onto the market.
โก WEKApod delivers 9,000 concurrent AI jobs and 880 GB/s of throughput from just 16U of storage.
Thatโs 80% more jobs than our 2025 submission.
See the full results โฌ๏ธ https://t.co/v2bx9AWqz6
๐ธ When GPUs wait on data, compute goes to waste.
WEKA NeuralMesh helps NCHC boost LLM training throughput, GPU utilization, and inference startup.
See how โฌ๏ธ https://t.co/GqlvOMP3vE
Training is about speed. Inference is about economics.
WEKAโs @liranzvibel and @stevehou of @silicondata recently connected to explore why inference is reshaping #neoclouds.
Catch the full conversation ๐ https://t.co/XUvZApNYaT
๐ช A bit of #Diwali joy from our Campbell office!
Some of our team came together to celebrate the Festival of Lights.
Happy Diwali to everyone who celebrated! Wishing you a year ahead filled with light, happiness, and prosperity. โจ
#WEKALife
๐ฐ๐ท The WEKA K-Team really enjoyed this event and the chance to take part in Koreaโs AI community coming together for AI Day Seoul 2026!
Learn more: https://t.co/QqtblKHZt9
The "Home Team" as they are called, are one of the reasons Singapore is such a powerhouse of achievement. Amazing focus and capability. It is an honour to work with them!
โก HTX, @Singtel RE:AI & WEKA achieved 99.49% GPU utilization for sovereign Physical AI, with multitenancy enabled.
See how ๐ https://t.co/oj8v5WvnCV
With David Jones in receivership and its future uncertain, it was ironic to see they are still the feature ad (today Nov 7) in the Qantas inflight captive portal. Might not be the best brand to promote anymore.
๐๏ธ How far away does your storage feel when your GPUs are waiting on data?
Legacy architectures werenโt built for AI-scale demand. WEKA NeuralMesh was.
โก๏ธ Take a closer look: https://t.co/qCTLNV8Aky
The JW Marriott Singapore has a ladder in the lobby that is not actually a ladder. It has the appearance of a ladder, but cannot do the job. like some vendors when they claim they have a parallel file system but actually they just have a 1990s architecture that they've warmed up
โฐ Just 2 weeks until #SC26 takes over Chicago!
We're gearing up for a great week of AI and HPC innovation, conversations with industry leaders, and a few fun surprises along the way.
๐ Get the full details: https://t.co/0CdckHelXO