Lablup Inc.

We brought a model server for your DGX Spark and MacBook. #MLXcel - The easiest way to run safetensor models on your DGX Spark and MacBook. Interested? Visit Booth #243 at GTC 2026. Learn more about Backend.AI:GO 📎 go.backend.ai

We're thrilled to announce that @lablupinc is officially part of the VAST Cosmos Partner Program, @VAST_Data 's unified global partner ecosystem built to accelerate AI infrastructure at scale. #Lablup #VAST_Data #VAST_COSMOS

🤝 Excited to make a MOU with PTC System under the newly launched Korea–Singapore AI Alliance! Together, we’re driving scalable #AI adoption across Asia. Let's advance AI infrastructure, HPC innovation, and ecosystem development across both nations. #Partnership #BackendDotAI

🚀 Excited to announce MOU with Knovel Engineering under the Korea–Singapore AI Alliance! Together, Lablup and Knovel will drive scalable #AI deployment, combining Backend.AI with Knovel’s AI expertise to accelerate real-world adoption. #Partnership #BackendDotAI

🚀Our CTO joins @Vultr on stage at #GTC26! See how Backend.AI delivers up to 4× higher GPU utilization with container-level GPU virtualization & the Sokovan engine — powering efficient multi-tenant clusters. Come talk #AI with us! #NVIDIA #BackendDotAI #Vultr #Lablup

🤷What happens when you run 60 B200 nodes for 73 days? Join our CEO's theater talk at #GTC2026 to explore the insights and lessons from real-world AI infrastructure operations. nvidia.com/gtc/session-ca…

Silver Sponsor, third year running. #Lablup coming to San Jose! 🚀 @NVIDIAGTC #Lablup #NVIDIA #BackendDotAI #GTC2026

We're at MWC Barcelona 2026! Backend.AI - One platform for all your AI accelerators Backend.AI:GO - Run sLMs offline on your desktop, connect multiple devices to run bigger models. Hall 7, 7A62-01 tally.so/r/yPJ5px #MWC26 #BackendDotAI #Lablup

⚡ Excited to announce that @lablupinc has integrated @FuriosaAI's RNGD into our infrastructure! Backend.AI maximizes RNGD workloads for low-latency inference and high performance-per-watt — together enabling next‑gen AI acceleration.

Here's a piece from @lablupinc on monitoring, failure handling, and recovery of 500+ GPU clusters. Curious about the uncomfortable realities of large-scale AI infra at hardware/software limits? Good place to start. Read the breakdown: backend.ai/blog/2026-02-l…

NVIDIA DGX Spark leverages Blackwell architecture via the GB10 Superchip for desktop AI supercomputing. Our analysis highlights how its SM12x compute units differ from SM100 in tensor core instructions like tcgen05 and TMEM support. Read the breakdown: bnd.ai/spark-analysis

We’re proud to announce that we’re now a Participating Sponsor of @ThePSF! From our first @PyConKR to managing over 16K GPUs across 110+ sites today, the Python ecosystem has been at the core of everything we build. Learn more in our newsroom: backend.ai/blog/2026-02-l…

11 years ago, our last post here said "Hello Twitter!" Sorry about the gap, we were building something big. Let's fix that. Nice to meet you (again).

Hello Twitter! #myfirstTweet