Every major technological transition in American history — the railroad, the interstate highway system, the internet — succeeded because government and the private sector built it together. Not because government directed every engineering decision, but because elected leaders understood what was at stake, asked the right questions, and ensured that public interests were part of the foundation.

We are at that moment again. Machine Intelligence is not a product category or a software trend. It is becoming a foundational layer of computing infrastructure — not yet as settled or consequential as the electrical grid, but heading in that direction as more of the economy comes to depend on it. Like earlier infrastructure transitions, the decisions made now about how it is built, who controls it, and what standards govern it will be harder to reverse the longer they wait.

The MIOS initiative is proposing an open standard for the operating system layer of machine intelligence. We believe the United States has a strong interest in helping lead that standard — not ceding it to fragmented proprietary platforms or to competitors moving without public accountability. We also want to be direct that this is a competitive, not empty, field: hyperscalers already run proprietary systems solving parts of this problem, and the Linux Foundation's UXL Foundation is already building pieces of the open hardware standard MIOS would rely on. Government's role isn't to pick MIOS as a winner in advance — it's to have informed people in the room while these standards are being set.

That kind of leadership does not happen without government engagement. It requires policymakers who understand the stakes, champions at every level who can articulate why this matters to their communities, and officials who are willing to ask hard questions of the technology sector on behalf of the public.

What Is MIOS, in Plain Terms

Today's most powerful AI systems run on top of operating systems that were designed in the 1980s — systems built for word processors and email, not for the kind of computation that drives modern machine intelligence. This mismatch wastes enormous amounts of energy, slows development, and forces every company building AI to independently work around the same set of outdated constraints.

MIOS proposes to fix that at the foundation. A purpose-built operating system for machine intelligence — developed as an open standard by a consortium of leading technology companies and research universities — would reduce energy consumption, accelerate development, and create a shared, accountable infrastructure layer that the entire industry can build on.

Think of it as proposing a new kind of road. Right now, every technology company is paving its own path through the same terrain. MIOS proposes to build a common highway — open, standardized, and accessible — so that competition can happen on top of it rather than underneath it.

The governance of foundational technology infrastructure is not exclusively a federal question. The effects of machine intelligence — on local employment, on public services, on education, on healthcare delivery, on civic infrastructure — are felt at every level of government. And so the conversation must happen at every level.

We are not asking officials to become technical experts. We are asking them to be engaged, informed, and present — to represent the interests of their constituents in a conversation that is happening whether or not government participates.

The MIOS initiative does not arrive with a pre-written legislative agenda. We are not asking government to adopt a position before the conversation has begun. What we are identifying is the set of policy questions where government engagement is most urgently needed — questions that will be answered by someone, and that should be answered with public input.

We recognize that inviting government into a technical conversation carries a responsibility to make that conversation accessible, honest, and genuinely two-way. The MIOS initiative makes the following commitments to every government official and agency that chooses to engage.

In keeping with the transparency commitment above, we commissioned an independent technical review of the MIOS proposal rather than only publishing our own account of it. It confirmed the core technical diagnosis, and also identified real gaps: our original roadmap's timeline was too aggressive relative to historical precedent for infrastructure standards, our security-verification budget was underscoped, and we had not adequately addressed the Linux Foundation's overlapping UXL Foundation effort or the incentives of incumbent hardware vendors. We've revised the white paper to reflect those findings directly rather than downplay them.

Read the full independent assessment →