NSF Just Did Something Important

And so can you: how to reach out to Convergent about X-Labs

Last week, NSF announced $1.5 billion for the NSF X-Labs initiative — independent, mission-driven research teams pursuing platform technologies that traditional institutions aren’t built to tackle. The first two topics are Scientific Instrumentation for Sensing and Imaging, and Quantum Systems: Interconnects and Integrated Photonics.

A few years ago, Caleb Watney at IFP published a high-level proposal for exactly this kind of program. He called them X-Labs and described four basic award types. The X02 category — “focused nonprofits building critical tooling with startup-like agility” — described, almost exactly, what we’d call a Focused Research Organization (FRO).

The Future of Focused Research Organizations:

The fact that NSF not only ran with the concept but committed $1.5B to it is a bold and exciting opportunity. We want to find the scientists and entrepreneurs who want to make the most of this moment.

What are X-Labs?

NSF X-Labs are independent teams with operational autonomy and milestone-based funding building platform technologies that existing academic structures can’t support. They’re not conceptualized open-ended research centers, individual PI or consortium grants, nor are they “centers” at existing academic institutions. Instead, they’re full-time teams in a dedicated, independent organization, built around a concrete technological target.

One interpretation? That’s a Focused Research Organization.

At Convergent, we build FROs to solve scientific bottlenecks that are too big for a single lab, too coordination-intensive for a loose collaboration, and not profitable enough for venture capital. The FRO team owns a specific deliverable — a tool, dataset, or platform — and pursues it with startup-like speed. The goal is to improve, by orders of magnitude, some underlying capability that reshapes what’s possible for an entire fieldoof science or technology.

The NSF X-Labs reflect these key characteristics, meaning that they are the first major US federal vehicle for FROs and similar organizations. We’re excited about this, as we’ve been experimenting with and prototyping this model since 2021, launching 13 FROs to date.

The Sensing & Imaging Topic Is One We’ve Been Thinking About

In our piece on Prioritizing Fundamental Capabilities, we argued that one of the most important things we can do right now is complement AI, building the measurement infrastructure that AI cannot substitute for. Even superhuman reasoning can’t measure contingent facts about the universe. It can’t substitute for actually seeing inside a living cell or brain. It can’t resolve protein structures that lie outside the distribution of what has been measured.

The bottlenecks here are physical, and they’ve been accumulating. We’ve catalogued some of these bottlenecks on our Gap Map. Here are a few that seem to sit directly in the X-Labs sensing and imaging territory:

• We can’t see deeply into living tissue. Light scatters. The deeper you go, the worse the resolution — which is why we can’t non-invasively map neural activity at single-cell resolution throughout a living brain, and why most diagnostics still require invasive procedures. Anti-scattering methods — like opto-acoustic imaging, time reversed light, computational wavefront correction — are promising, but nobody has built the coordinated platforms and large-scale microchips to push them over the hump.

• High-resolution live cell imaging destroys the sample. You can have nanoscale resolution or you can have a living cell. Not both, not yet. Quantum non-demolition imaging approaches — ghost imaging, quantum electron microscopy — are early-stage but hold promise. Raman microscopy might one day map the transcriptome of a cell just by looking at it. The path from proof-of-concept to instrument platform requires the kind of sustained engineering effort that a single academic lab or loose consortium/center can’t provide.

• We can’t take movies of brain computation. We can record from a few neurons or get population-level blurs. Single-neuron, millisecond resolution across large networks in a living brain is out of reach with current tools. In-vivo connectomics, new fast-scanning microscopy methods, optical clearing of live tissue — these are all emerging partial answers to the same problem.

• Material structures and properties at the atomic scale remain opaque. Neutron microscopy offers deep penetration and sensitivity to light elements, but accessible, high-resolution neutron imaging doesn’t exist as a practical research tool. DARPA ran the ICONS program. Progress was made. Nobody closed the loop. Prototype microscopies are starting to image the complex electronic states of quantum materials but they aren’t yet scalable.

• Key biomolecules resist structure analysis. AlphaFold changed what’s possible for soluble proteins. Membrane proteins — which make up a huge fraction of drug targets — often remain difficult. Cryo-EM struggles with small proteins. Experimental structure analysis is still a bottleneck for a large fraction of the proteome, not to mention the glycome, lipidome or other biomolecular families. Meanwhile, microscopes are starting to see protein shapes directly.

• Quantum effects in biology are unmeasured. Theory has outpaced measurement. Where quantum coherence actually plays a functional role in biological systems remains genuinely unknown because we lack the instrumentation to find out. This is beginning to change.

• We can’t monitor human physiology longitudinally and noninvasively. Breath analysis, wearables that capture high-dimensional physiology — none of these have reached the point where repeated, high-resolution physiological data is routinely collectable outside a clinical setting. The data that would transform our understanding of human health over time doesn’t exist, largely because the sampling tools don’t.

Each of these is a critical bottleneck — solve it and you unlock downstream progress across multiple fields. But to do so you need a concerted, focused, cross-disciplinary organization. That’s what makes these bottlenecks — and a number of other similar problems — FRO-shaped. And that’s what also makes them X-Lab-shaped.

Here’s how the NSF put it:

Every revolution in science has been preceded by a revolution in what we can measure, from the telescope to modern Magnetic Resonance Imaging (MRI) machines. Today, the frontier is starved for radically new modalities for sensing and imaging. We cannot watch a non-crystalline enzyme work at atomic resolution, probe the full dynamics of a working synapse, or identify the most reactive surface defect structures on advanced catalytic materials. NSF X-Labs in this Topic will target specific platform technologies in sensing, imaging and supporting technologies that will form the basis for revolutionary new capabilities in scientific discovery and technology sectors. Teams might, for example, draw on quantum sensing, artificial intelligence (AI)-driven computational imaging, adaptive AI-based sensing algorithms, and/or entirely new modalities to redefine what we consider knowable. Examples of relevant, currently unmet R&D challenges may include, but are not limited to: detection of molecular-scale single-reaction events across timescales of femtoseconds to seconds; MRI-free deep-tissue imaging; non-destructive biomolecule microscopy at exquisite resolution; high-sensitivity quantum sensors suitable for operation in a variety of environments; instruments intentionally engineered for next-generation AI training pipelines; and sensors to resolve whole-brain activity at cellular resolution across long timescales.

In our next post, we’ll say more about the bottlenecks we’re learning about around the quantum photonics and interconnect topic.

We’re Looking for People Working on This

We’re gearing up to work with some of the world’s most technically and operationally talented and ambitious individuals and teams – those who are already in these spaces or who’ve been thinking about them but haven’t found the right vehicle yet.

The people we’re looking for aren’t necessarily thinking “I want to start an FRO.” Instead, they’re probably thinking: “there’s a measurement capability that doesn’t exist and should, and I think I know how to build it.” They might have written a technical roadmap that nobody funded. They may have tried to assemble a team inside an academic setting and hit the ceiling on the roles and modes of teamwork that incentive structure readily supports. They probably have a very specific build target in mind and strong opinions about why the status quo approaches are insufficient.

That profile — a specific, transformative bottleneck, a concrete build target, forward-looking roadmaps grounded in deep technical knowledge, and a frustration with existing structures that make them want to go full-time on operationally building a fast-moving, laser-focused professional team — is what we look for in FRO founders. Strong founders come from all over: recent PhDs, senior academics, industry engineers, former ARPA program managers. What unifies them is the drive to build something, not just study something.

If that sounds like you, or someone you know, we want to hear from you. Contact us by completing this short form.

We want to support the X-Labs program as the initiative develops — helping nucleate teams, helping them develop ideas into fundable proposals, and building the organizational infrastructure to help the best ones launch and succeed.

There’s a lot of work to do. Initial applications to NSF are due in mid-July. Let’s talk soon.