06/15/2026
What if your organization could cut its AI token costs by up to 70%?
That's the promise of Hybrid AI β and it's why the Princeton IT Services team just completed an intensive Microsoft AI training, hosted by Microsoft, Qualcomm, and TD SYNNEX at the Microsoft Innovation Hub!
Our technical experts went deep on:
π· Hybrid AI β intelligently splitting workloads between on-device NPUs and cloud frontier models
π· Windows AI Foundry on Copilot+ PCs and Microsoft Surface devices
π· Local AI inferencing with Phi Silica and Foundry Local
π· Qualcomm Snapdragon X1 (45 TOPS) and X2 (80 TOPS) NPUs powering on-device AI
π· Optimizing workloads across CPU, GPU, and NPU β from Copilot+ PCs to edge servers
The big takeaway: Not every AI workload needs a frontier model in the cloud. With NPUs now delivering up to 80 TOPS on-device, businesses can run the right tasks locally β dramatically reducing token spend while improving latency and data privacy.
A huge thank you to the incredible partners who made this high-impact training session possible:
Microsoft: Laura Osborn , Neil Misak , Rebecca Garrity
Qualcomm: Craig Tellalian , Michael Liwanag
Teknikos: Jonathan Khoo , Jason Hartman
TD SYNNEX: Russ Seaman , Ken Lince, Jr. , Tj Turner , Kim Dillard
At Princeton IT Services, we continuously invest in upskilling our team so our clients always stay ahead of the curve β not behind it.
π‘ Wondering how much your organization could save on AI spend?
Reach out to Mahendra Kumar Keshavpatnam to explore how Princeton IT Services can help you build a smarter, more cost-effective AI strategy.
β β β β β β β β β