At the Teratec Forum, Kacem Lounissi, Head of AI and MLOps at Sesterce, joined the roundtable on "Infrastructures and Data for AI Factories". He shared Sesterce’s approach to building scalable, efficient, and sovereign AI platforms. Thanks to Emmanuel Duteil for a

Just dropped: 8x B200 bare-metal nodes — now live in Europe. 🔹1440 GB GPU vRAM 🔹 2304 GB system RAM 🔹 No virtualization. Full control. EU-hosted. Built for: ✅ Training reasoning-first LLMs ✅ Scaling inference at speed ✅ Distributed AI workloads → Live now on Sesterce

Just 1 week before #GTCParis — we’re unlocking something big: Cluster LUNA is now live — 128× NVIDIA H200 GPUs, 3 TB RAM/node, 500 TB VΛST Data storage. Hosted in Paris. Sovereign. SLURM-ready. No egress fees. 🔹 $1.99/h — GTC Paris exclusive 🔹 Only 1 cluster available.

If you're building any of these, you know that time-to-compute is everything. That's why we've opened access to GH200 96GB GPUs at just $4.65/hour — fully on-demand. No lock-in. No waiting. Just direct access to one of the most powerful AI chips out there — with 96GB of unified

Got a bold idea for AI? In 5 days at #GTC25, we’re kicking off the #LUNAChallenge — your shot to win a share of $20,000 in AI prizes. No code needed. Just vision. Stay tuned. #PrivateAI #H200 #GTC25 #Sesterce

🇫🇷 Today at #F5AppWorld Paris, our CEO Youssef El Manssouri joined a powerful roundtable on scaling secure & sovereign AI infrastructure. 🔹 With Ahmed Guetari F5, Nathaniel Ives NVIDIA 🎙️ Moderated by Alix Leconte Europe’s AI must run on trusted, high-performance foundations. #Sesterce

🚨 New Journal du Net Benchmark: Sesterce ranks #1 in AI Cloud pricing Lowest public prices on the market: 🔹H100: $1.79/hr 🔹H200: $2.48/hr Cheaper than AWS, GCP, CoreWeave. 🔹 Sovereign infra 🔹 PUE 1.1 datacenters 🔹 Scale from 1 to 15,000 GPUs ➡️ journaldunet.com/cloud/1542459-…

Choosing the right compute instance shouldn’t be a guessing game. We broke it down so you can pick based on what you’re training, generating or deploying: 🔹Training vs. inference 🔹LLMs vs. vision 🔹H100 vs. GB200 ➡️ Tutorial : bit.ly/3TJUhQ5 What tutorial should we

“France has 5 GW of unused energy — let’s build sovereign AI datacenters!” At Sesterce, we believe France can be Europe’s AI powerhouse. By 2030, AI needs 100 GW — France must act now. 🎥 Watch Youssef El Manssouri on why France could lead Europe’s AI future: bit.ly/46bmqr1 #AI

Want to serve a Hugging Face model like an API? We wrote a step-by-step guide to do it with vLLM—fast, efficient, and OpenAI-compatible. 🔹 Works great on Sesterce 🔹 Easy to deploy 🔹 Ready for production ➡️ Full tutorial → bit.ly/45gHk6Y

Already told a friend about Sesterce? Might as well earn from it. Every user gets a referral link. Share it → earn 3% commission on every transaction your contacts make. No catch. Just clean revenue for helping others scale GPU infra fast (and sovereign). ➡️

Generic cloud slows research. Competitive AI R&D needs Custom GPU Clusters: ✅ InfiniBand/NVLink ✅ Sovereign hosting ✅ Fleet observability This is how AI innovation scales: bit.ly/41fxcc7

Open models, without lock-in. Deploy Pixtral, Llama 4, Qwen 3 on: ✅ Sovereign EU infra ✅ GPU-accelerated endpoints ✅ Full compliance, zero setup Start building ➡️ bit.ly/4mdh7MV #AIInference #SovereignAI #Llama4 #Qwen3 #Pixtral

Deployed in France. Live in seconds. No fine print, just sovereign GPU clusters: ✅ H200, H100, L40, L4 ✅ Regions: Paris & Marseille ✅ Ubuntu-based envs ✅ No-commit pricing Born in 🇫🇷. Built for global AI. ➡️ bit.ly/3H8ye30 #GPUCloud #SovereignAI #HPC #PrivateAI

Infra should be fast. Not buried in tabs. Sesterce Cloud is 100% CLI-native. No dashboards. No dropdowns. Just: ✅ GPU VMs & Bare Metal ✅ Region & type flags ✅ Script & CI/CD ready 🖥️ sesterce create --region=par4 --gpu=H200 --count=8 ➡️ bit.ly/4lKSbLp

Training gets the glory. But inference runs the show. Every chat, every recommendation, every AI task = inference. Live. Continuous. Revenue-critical. Still scaling like it's 2021? You’ll hit latency walls fast. Inference is your factory. Architect like it. #AIInfra

What an amazing few days we had in San Francisco! Sesterce team was right in the thick of things at the PyTorch Summit (Oct 22-23), and let me tell you, the energy was absolutely buzzing. The big takeaway? The future of AI isn’t just about creating larger models; it’s about

The wait is over! We’re so excited to announce the GPU MODE x @NVIDIA kernel optimization competition for NVFP4 kernels on Blackwell B200s! We will be awarding NVIDIA DGX Spark’s & RTX 50XX series GPUs for individual rankings on each problem, as well as a Dell Pro Max with