Microsoft debuts Surface RTX Spark Dev Box to run large AI models without cloud costs
THE SO WHAT
A compact RTX Spark Dev Box that runs large models locally is a direct attack on per-token cloud economics — inference is being reclassified as capex, not opex. If your business model depends on metered API usage, start planning for customers to internalize the heaviest workloads on their own hardware.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIGoogle’s Phone app will tell you if a scammer is impersonating one of your contacts
Carrier identity is now an AI problem—phone apps are becoming real-time fraud filters, not just dialers. If you run consumer comms or fintech, assume your users are being targeted by AI impersonation and build in-call verification and out-of-band checks as a first-class feature.
Applied AIEnterprise AI agents keep creating data silos. Microsoft's Build answer is Microsoft IQ and Rayfin.
The agent sprawl problem is now explicit—every uncoupled agent is a new shadow app and data silo. If you’re greenlighting agent projects without a unifying data and policy layer like Microsoft IQ/Rayfin, you’re rebuilding the pre-SaaS mess with better autocomplete.
They tricked Meta's chatbot to hack into Instagram accounts. It's a warning about how much we trust AI.
Users now treat chatbots like trusted ops staff—and attackers are treating them like a new social engineering surface. If your AI assistant can touch auth, payments, or account settings, you need the same red-teaming and least-privilege controls you’d apply to a junior SRE.
Applied AIMicrosoft announces new on-device AI updates for Edge: a dev preview of a new SLM called Aion-1.0-Instruct, Language Detector and Translator APIs, and more (Paul Thurrott/Thurrott)
On-device SLMs in Edge turn the browser into an inference runtime, not just a window to cloud models. If you build extensions or web apps, assume a local model is available and start shifting latency-sensitive, privacy-sensitive logic client-side.