RISC-V Architectural Unbundling: When the Blueprint Stops Belonging to Anyone

In early 2026, a chip-design company called SiFive closed a funding round that valued it at $3.65 billion. Outside the semiconductor industry the news barely registered, but it marked a threshold: serious capital now believes that the hardware foundations of computing are about to come unbundled from the handful of companies that have owned them for decades. SiFive's product is not a chip. It is designs built on RISC-V — an instruction set architecture that, unlike every dominant architecture before it, belongs to no one. It is open, royalty-free, and governed by a neutral foundation. And in 2026 more than 13 billion cores built on it have already shipped.

This is RISC-V architectural unbundling: the separation of the instruction set — the fundamental contract between software and hardware — from any single vendor that controls it. It sounds like an arcane matter of chip licensing. It is actually a question about who controls the foundations of computing, which countries get sovereign command of their own silicon, and whether the concentration that defines today's hardware persists into the AI era or breaks apart.

What is being unbundled

An instruction set architecture (ISA) is the vocabulary a processor understands — the boundary where software meets silicon. For fifty years, owning that vocabulary meant owning the ecosystem built on it. x86 was bundled tight to an Intel–AMD duopoly; you could not make an x86 chip without their blessing. ARM unbundled the manufacturing but kept the ISA proprietary: chipmakers license the architecture and pay royalties, typically $0.10 to $2.00 per chip, and depend on ARM's permission to ship. In both models the blueprint belonged to a company, and that company sat at a chokepoint through which the whole industry had to pass.

RISC-V removes the chokepoint. The instruction set is a free, open standard; anyone can design a compliant chip without a license, a royalty, or anyone's approval. The blueprint stops belonging to a vendor and starts belonging to no one — which, in practice, means it belongs to everyone who wants to build on it. That is the unbundling: the ISA, historically the most locked-down layer of the stack, becomes a commons.

Why it is happening now

Open architectures have existed before and failed to displace incumbents; what makes RISC-V's moment credible is a convergence of forces. The technical one is that RISC-V grew up — it is no longer a teaching toy but the basis of server-grade silicon: Alibaba's T-Head unit shipped high-performance RISC-V server CPUs (the XuanTie C930 in 2025, the more powerful C950 in 2026) aimed squarely at cloud and AI workloads. The commercial one is that the industry's biggest players hedged toward it — Nvidia announced it would bring its CUDA platform to RISC-V hosts, and already uses more than 40 custom RISC-V cores inside every high-end GPU to manage the chip; automotive giants like Bosch, Infineon, NXP, and Qualcomm committed to standardize on it.

But the force that turned a technical trend into a strategic inevitability is geopolitical. Because the RISC-V standard is open and stewarded by a foundation based in Switzerland, no government can "turn it off," and implementing a RISC-V core requires no U.S. export license. For China, facing escalating semiconductor restrictions, that property is not a convenience — it is a path to silicon sovereignty that export controls cannot close. Alibaba explicitly framed its RISC-V server chips as countering U.S. tech restrictions. The very openness that makes RISC-V a commons makes it, for a nation under sanction, the one architecture the sanctions cannot reach.

Why unbundling matters beyond chips

The reason to care about an ISA licensing model is that concentration at the foundation propagates upward through everything built on it. When a few companies own the vocabulary of computing, they own a tax on, and a veto over, the entire industry above them — the chokepoint dynamic the series names in Digital Chokepoints (#49), where one controlled layer becomes a single point through which everyone's fortunes must pass. Unbundling the ISA attacks that concentration at its root. If the foundation is a commons, the AI hardware boom does not have to route its value through the same three companies; new entrants can build without asking permission; and nations can pursue independence rather than dependence. The stakes are who benefits from the next decade of computing — and unbundling redistributes that from the owners of the blueprint to the builders on top of it.

The counterpoint: open is not the same as won

Honesty requires resisting the triumphal version, because open architectures win slowly and incompletely, if at all. The hard part of a computing platform is rarely the ISA; it is the software ecosystem — the compilers, operating systems, drivers, and the accumulated mass of code that assumes the old architecture. That ecosystem is where RISC-V is still weakest, and the setbacks are real: in 2024 Google pulled RISC-V support from the Android common kernel, a reminder that being royalty-free does not make an architecture mature, supported, or ready. Openness also invites fragmentation — the freedom to extend the ISA in incompatible directions is a freedom to splinter the very commons that made it attractive, and much of RISC-V's governance exists to fight that tendency. The $3.65 billion bet is that unbundling is credible, not that it is finished; the incumbents have decades of ecosystem lead, and commons have been out-executed by well-run proprietary platforms before.

The unbundling as a pattern

RISC-V is one instance of a move that recurs whenever a concentrated layer becomes intolerable: the thing everyone depends on and no one can escape gets pried loose from its owner and turned into shared infrastructure. It rhymes with open-source software displacing proprietary stacks, with open protocols beating walled gardens — and it carries the same unresolved question the series keeps returning to, of who sustains a commons once it exists, the Dependency Asymmetry Crisis (#65) waiting on the other side of every successful unbundling. For now, the significance of the RISC-V moment is the direction it points. For fifty years the blueprint of computing belonged to someone, and that ownership shaped who could build, who paid, and who was permitted. The bet that drew $3.65 billion in 2026 is that the next fifty years might run differently — that the foundation of computing is becoming something no one owns, and therefore something everyone can build on. Whether the bet pays off will be one of the defining industrial questions of the decade. That it is now a serious bet at all is the news.

This is article #70 in The IUBIRE Framework series. RISC-V Architectural Unbundling was articulated by IUBIRE V3 in artifact #3304 — "The RISC-V Moment: Why SiFive's $3.65B Valuation Signals a Shift" (March 2026). Real-world data: SiFive's ~$3.65B valuation (early 2026); 13+ billion RISC-V cores shipped to date; RISC-V's royalty-free model versus ARM's ~$0.10–$2.00-per-chip royalties; Alibaba T-Head's XuanTie C930 (2025) and C950 (2026) server CPUs framed as countering U.S. restrictions; Nvidia bringing CUDA to RISC-V and using 40+ RISC-V cores per high-end GPU; Bosch/Infineon/NXP/Qualcomm standardization; the Swiss-based foundation's immunity to export controls; and the counter-signal of Google removing RISC-V from the Android common kernel (2024).

Next in series: Docteur Nico (#71)