Why location decisions matter more in hardware than software
Software startups can be built almost anywhere. The dependencies are network connections, code repositories, and the ability to recruit engineers who can work with standard tools on standard hardware. Deep-tech hardware is different. The dependencies include fabrication partners, metrology equipment, optics suppliers, and crucially, a local community of people who understand the physics. Getting those dependencies wrong adds months of lead time and friction to every iteration.
When we started Great Sky in 2023, the choice of Boulder was not just about where the team happened to be living. It was a deliberate assessment of what the next several years of work would require and where those requirements could be met.
Proximity to NIST Boulder
The National Institute of Standards and Technology has a major research campus in Boulder, with a significant effort in photonics, optical frequency standards, and quantum optics. NIST Boulder is not a commercial entity and we are not describing any formal partnership. But a research institution of that concentration creates a talent pool. Engineers and scientists who trained at NIST, worked there, or collaborated with NIST researchers while at CU Boulder make up a meaningful fraction of the optics-competent population in this area.
That is not a minor convenience. When you are trying to hire a process engineer who understands silicon photonic waveguide fabrication and has operated PECVD equipment, the number of such people in the country is small. Boulder has an above-average concentration of them relative to its population. The same holds for optical test engineers, cryogenic systems people (adjacent to quantum photonics), and precision measurement specialists.
We are a small team. We cannot offer equity stakes that would lure someone from a stable NIST position. But we can offer an environment where their background is understood and valued, surrounded by colleagues who speak the same technical language. That matters to the kind of scientist or engineer who wants to work on deep-tech hardware.
CU Boulder and the academic pipeline
The University of Colorado at Boulder has an active photonics research program across its electrical engineering and physics departments. The work ranges from fiber amplifier development to semiconductor photonic devices to optical coherence tomography systems. Graduate students from these programs represent the most direct talent pipeline for a company doing what we do.
Being in the same city as the university is not a guarantee of anything. We have to be interesting enough for researchers finishing PhDs to consider joining a small team over the traditional options. But the friction is lower. A student can attend a public seminar we give, visit the office after, and have a real conversation. That kind of informal contact is hard to replicate when you are a six-hour flight away from your talent pipeline.
We are not claiming an exclusive arrangement with CU or any formal industry partnership. What we have is the advantage of physical proximity and the ability to participate in the local research community in a way that keeps our name visible to the people most likely to want to work on photonic inference hardware.
Hardware iteration rhythm in a landlocked city
One objection to Boulder as a hardware location is the absence of a major semiconductor foundry nearby. That is a real constraint. We tape out through commercial silicon photonics foundries that are not in Colorado. Lead times are what they are, measured in weeks to months, regardless of where we are located. A company in San Jose or Boston would have the same foundry situation for silicon photonic processes, because the relevant foundries are in the US Pacific Northwest, Europe, and Taiwan.
What does exist locally is a reasonable set of optical test equipment suppliers, fiber assembly houses, and precision machining shops. This is more important than it might appear. A significant fraction of early-stage hardware iteration involves building test fixtures, modifying probe stations, and customizing fiber coupling setups. These are mechanical and optical fabrication tasks that are faster and cheaper to iterate on when you have local machine shops and optics vendors.
There is also a less tangible benefit: the software and hardware talent pool in Boulder includes people who have worked on physically demanding products, not just web applications. The engineering culture here has been shaped by the aerospace and defense companies that have been in the area for decades. That translates into engineers who think carefully about tolerances, who do not assume cloud deployments and restarts fix everything, and who understand that hardware does not have a software rollback.
The altitude, actually
Boulder sits at roughly 1655 m elevation. This is irrelevant to our product and to most of our work. It does, however, affect one practical consideration: any hardware we build that is intended for deployment at sea-level data centers needs to be tested at operating conditions, not just at altitude. Convection cooling behavior changes with air density. Our test setup includes forced-air cooling calibrated to simulate sea-level conditions rather than our ambient environment. This is a small, solvable problem, but it is the kind of thing that only becomes obvious once you are actually building the hardware.
What Boulder is not
We are not claiming Boulder is an established photonics hardware cluster on the order of the Research Triangle or the Massachusetts Route 128 corridor. Those regions have decades of compound semiconductor and photonics industry concentration. Boulder is smaller and younger as a hardware ecosystem.
What Boulder has is a specific combination: federal research lab concentration, a strong research university, a community of technically trained people who chose to be here for reasons unrelated to career advancement, and a lower cost of living than the Bay Area or Boston. For a small team doing early-stage deep-tech work, that combination is useful. We are not selling the city as a destination. We are saying that the decision to build here reflected a real analysis, and the analysis has held up so far.
If you are a photonics engineer or optical systems researcher thinking about what it looks like to work on this problem with us, the answer is: you would be in a place where the work is taken seriously and the physics is understood by your colleagues. That is the environment we have tried to build, and Boulder makes that easier than many alternatives.