@mor10web: Microsoft and NVIDIA are banking on local AI models being the Next Big Thing. And I think they're right on the money here, because of privacy and token cost. With protests against AI datacentres happening across the globe, AI token costs soaring into the millions for even small companies, and data privacy becoming front-of-mind for companies and individuals, the idea of cutting the cloud cord and running language models locally is becoming more and more appealing. The main blocker for this shift-to-local AI is hardware cost and availability. Local language models require dedicated memory and processing. Even the most overbuilt laptops on the market can barely run the larger models that start rivalling those from OpenAI and Anthropic, and dedicated AI inference machines start at over $4000 US. That's a huge cost, but considering many developers are spending several thousand per month on tokens, even that eye watering price tag starts making sense. It's with all of this as backdrop that new hardware news hit the mainstream this week: Between Nvidia's new RTX Spark (a tiny chip with enough power to run huge LLMs locally on small laptops) and Microsoft's new Surface computers including the Surface Laptop Ultra and the desktop AI powerhouse Surface RTX Dev Box, it is clear the two tech behemoths are reading the room and meeting a growing demand. Once these machines hit the market, I expect the rush for Mac Minis and Studios will dry up quickly, and there's a good chance people who up until now have sworn allegiance to Apple will see a financial and performance gain in shifting from Mac to PC! The good news is you don't have to wait, or to buy a new computer, to run local models. Even my $50 Rasberry Pi can run some smaller models (slowly, and with hilariously bizarre results), and the free open AI models are getting faster and better every day. I think we're headed towards a future where the big AI companies have to make some pretty compelling arguments for people to go out and pay through the nose for tokens when they can run models at home for free. We shall see! Are you using local models? What are your experiences? Comments are open, and I'm curious to hear your experiences. Drop a line, create a video of your own, and let's figure out this AI stuff together! -- Links: Nvidia RTX Spark: https://nvidianews.nvidia.com/news/nvidia-microsoft-windows-pcs-agents-rtx-spark Microsoft Surface RTX Dev Box: https://nvidianews.nvidia.com/news/nvidia-microsoft-windows-pcs-agents-rtx-spark Microsoft Surface Laptop Ultra: https://www.theverge.com/tech/941600/microsoft-surface-laptop-ultra-dev-box-hands-on LM Studio: https://lmstudio.ai/ Ollama: https://ollama.com/ Hugging Face: https://huggingface.co/ #mortenexplains @surface @vscode #rtx #ollama #lmstudio

Morten of the North 🇨🇦🇳🇴

Open In TikTok:

Region: CA

Thursday 04 June 2026 00:00:24 GMT