Intel has quietly discontinued its controversial Intel On Demand initiative — also known as Software Defined Silicon (SDSi) — after the GitHub repository supporting the program was archived in November 2025 and material related to it was removed from Intel’s website, signaling the end of development and broader abandonment of the model that let customers pay extra to activate hardware features already baked into select Xeon processors. Reports indicate that Intel On Demand was originally positioned as a flexible way for enterprise customers to enable advanced accelerators — such as secure enclaves, load balancing, data streaming, and analytics accelerators — via a one-time payment or a consumption-based license, but widespread criticism over the idea of selling hardware capabilities already physically present but disabled by default appears to have dampened adoption and interest. Coverage from multiple outlets, including Tom’s Hardware and Phoronix, confirms the archiving of the SDSi repository on GitHub and the significant reduction of support and documentation by Intel, with the feature absent from recent product lines and unlikely to be part of future Xeon platforms. Some open-source observers also note this move is part of a broader pattern of Intel shelving a number of open-source projects in late 2025 and early 2026.
Sources
https://www.tomshardware.com/pc-components/cpus/intel-quietly-kills-controversial-software-defined-silicon-initiative-github-repository-was-archived-in-november-2025-allegedly-signaling-the-end-of-active-development
https://www.phoronix.com/news/Intel-On-Demand-SDSi-Sunset
https://www.technobezz.com/news/intel-discontinues-its-pay-to-unlock-cpu-feature-model
Key Takeaways
• Intel has discontinued its “Intel On Demand” / Software Defined Silicon initiative, archiving the supporting GitHub repository and removing most documentation.
• The model allowed customers to activate existing processor hardware features through additional payments, a concept that faced strong industry criticism for effectively charging twice for built-in hardware.
• The move aligns with a broader trend of Intel shelving certain open-source and experimental initiatives, and On Demand is not expected in upcoming Xeon product lines.
In-Depth
Intel’s recent discreet withdrawal of its Intel On Demand initiative — a program that let businesses activate extra silicon features on enterprise Xeon processors by paying for licenses — marks a quiet but significant pivot away from an experiment that stirred controversy from the start. First laid out as part of its Software Defined Silicon (SDSi) strategy, Intel On Demand was introduced with the promise of flexible hardware activation. Companies could buy server chips at a baseline price and then unlock specialized accelerators for workloads ranging from secure enclaves to load balancing and analytics via either one-time payments or usage-based licensing. In theory, this offered a way to shift cost from capital expenses into operational expenses, letting customers scale payments as needed. However, the industry reaction was skeptical at best: critics pointed out that these accelerator IP blocks were physically present in the silicon but disabled unless customers paid to enable them, a practice that many described as charging twice for the same hardware.
Over time, the initiative faded from public view, with Intel mentioning it less and less in product announcements and press materials. Sharp observers in the open-source community noticed that the SDSi GitHub repository — the hub for the software components required to support Intel On Demand — was archived in November 2025, a strong indicator that development had ceased. Following that, Intel cleaned up most of its supporting documentation on its site, leaving only legacy PDF files for those who dig deep enough. This suggested not just a pause but a wholesale retirement of the initiative.
Reports from respected tech outlets such as Tom’s Hardware and Phoronix underscore that Intel On Demand will not be part of future Xeon platforms and that the company is focusing elsewhere. Indeed, recent Xeon 6 and 600-series processors lack support for On Demand entirely, underscoring Intel’s refocus on core CPU performance rather than modular pay-to-unlock features.
What makes this development notable beyond Intel’s product roadmap is the broader context: the company has recently archived or shelved numerous open-source projects that it once maintained, signaling a change in its open-source engagement strategy. Whether this reflects a strategic retreat to reinforce strengths or a response to market and customer pushback, it’s clear that the controversial pay-to-unlock model of Intel On Demand didn’t gain enough traction to justify continued investment. For enterprise customers and server builders who once considered this licensing approach, the discontinuation removes an experimental — and for many, unwanted — option from Intel’s portfolio, leaving traditional hardware feature sets and licensing structures intact as the company moves forward.