Even people on #LinkedIn are now slowly waking up and starting to embrace local compute more.

So the statement from Jeff Bezos about wanting to kill local compute is once more of the lines of "If everything was hopeless they wouldn't have to try to discourage us so hard".

(Also local compute also includes any cloud that isn't theirs of course)

With things like:

* https://www.linkedin.com/posts/anthonymonthe_your-vms-costing-too-much-thats-normal-activity-7416781992987394049-qWQh
* https://www.linkedin.com/posts/uhitzel_i-have-thousands-of-dollars-in-unused-cloud-activity-7417428317772984320-Scxa
* ...

#cloud #IT #localcompute

Your VMs costing too much? That's normal... AI and its ravenous data centers are gobbling up global server capacity.  ‣ Dell/Lenovo announcing +15-25% on servers ‣ OVH expecting 5-10% VM hikes by… | Anthony Monthe

Your VMs costing too much? That's normal... AI and its ravenous data centers are gobbling up global server capacity.  ‣ Dell/Lenovo announcing +15-25% on servers ‣ OVH expecting 5-10% VM hikes by April ‣ DRAM's getting scarce, prices skyrocket, and boom… 💣 …your costs explode. 💥 AI isn't just revolutionizing your models; it's quietly breaking the hardware market…

Người dùng Reddit cập nhật hệ thống AI tại nhà sau tai nạn VRAM A6000 do làm mát kém. Đã chuyển sang nước làm mát (Bykski, ROMED8-2T+Epyc) dù làm mainboard hỏng sau rò rỉ. Nhiệt độ 45°C, chạy GLM 4.5 Air AWQ FP8. Mong chờ GLM 4.6 hoặc Mistral Large. Đang cân nhắc A2 (16GB) cho âm thanh/định nghĩa. #AI #LocalCompute #MáyTínhGiảiThuật #LàmMát #GPU #LocalLLaMA

https://www.reddit.com/r/LocalLLaMA/comments/1ozfsr7/local_rig_back_from_the_dead/

Plot twist: The most advanced AI transcription requires iPhone but acts like it's 1995. Everything local. Nothing shared. Privacy through time travel. https://apple.co/4nDlJvR #MeetingTranscription #iPhone #RetroTech #AIPrivacy #LocalCompute #Innovation #PrivacyFirst #iOS
Meeting Minutes: Pragmatic AI App - App Store

Download Meeting Minutes: Pragmatic AI by Nikola Knezevic on the App Store. See screenshots, ratings and reviews, user tips, and more games like Meeting Minutes…

App Store

🚨 NVIDIA launches AI-first desktops — no cloud needed.

💻 DGX Spark: 1 PFLOP, 128GB unified memory
💻 DGX Station: 20 PFLOPS, 784GB unified memory

✅ Native support for PyTorch, Jupyter, Ollama
✅ Same stack as DGX Cloud
✅ Ships with NIM microservices & 800Gb/s networking

Built for local, high-stakes AI: agentic models, sensitive data, and fast iteration.

Is cloud-only AI on its way out?

Video: NVIDIA

#NVIDIA #AI #EdgeComputing #AgenticAI #LocalCompute #TechNews #MastodonTech