Technology & Hardware · Featured Article
Sophia Space Raises $7M and Taps Apex for a 2027 Orbital Data Center Demo
Sophia Space has raised $7 million to bring data centers to orbit, betting that smarter thermal design — not brute force — is the key to processing satellite data in space. A SOOS software flight lands in fall 2026, with a TILE hardware demo on an Apex Nova bus in 2027.
By BlacKnight Space Labs, Space Industry Analysis · · 11 min read
- Sophia Space
- orbital data centers
- space-based compute
- edge computing
- TILE
- SOOS
- Apex
- Nova bus
- Kepler
- Rob DeMillo
- thermal management
- space economy
On June 23, 2026, Sophia Space confirmed it is finalizing a $7 million SAFE financing round, bringing the company's total funding to $22 million and setting up an ambitious two-flight roadmap to prove that data centers belong in orbit. The round is backed by EverGreen — the NVIDIA Alumni Investment Network — alongside SparkLabs Group and additional undisclosed investors. The capital will fund research and development, expand the engineering and sales teams, and stand up a manufacturing pipeline to scale production of the company's orbital compute hardware.
The Deal at a Glance
Sophia Space, led by CEO Rob DeMillo, is building the hardware and software to run generic, hostable data centers in space. The new $7 million round is structured as a SAFE — a common early-stage instrument that converts to equity in a future priced round — and lifts total funding to $22 million. Rather than spend the money on a single moonshot launch, Sophia is sequencing two focused demonstrations: a software validation flight in fall 2026, then a hardware demonstration in 2027.
The roadmap is deliberately de-risked. The first flight, in fall 2026, will validate SOOS — the Sophia Orbital Operating System — hosted on Kepler's in-space network, proving the software layer that manages compute in orbit before committing to bespoke hardware. What is learned on that mission then informs the second flight in 2027, when Sophia's flagship TILE compute hardware will fly on an Apex Nova satellite bus and interact with onboard sensors to run AI inference workloads such as image processing directly in space.
The Problem: Data-Rich, Decision-Poor
Modern satellites are extraordinary data collectors. A single Earth-observation spacecraft can gather terabytes — sometimes petabytes — of imagery and sensor readings. But collecting data is not the same as using it. As DeMillo describes the problem, satellites throw away the overwhelming majority of what they capture because they lack the onboard processing to analyze it and the bandwidth to send it all home. The scarce, expensive downlink becomes the chokepoint, and most of the value never makes it to the ground.
The fix is to move computation to where the data is born. If a satellite can analyze imagery on orbit — detecting a wildfire, identifying a ship, flagging a change — it can downlink a small, decision-grade answer instead of a mountain of raw pixels. That shift, from shipping data to shipping insight, is the thesis behind orbital compute, and it becomes more valuable as sensors proliferate and generate ever more data than any ground link can carry.
Sophia's Bet: Don't Fight Physics
The central engineering challenge of a data center in space is not the computing — it is getting rid of the heat the computing produces. In the vacuum of orbit there is no air to carry heat away; a spacecraft can only shed heat by radiating it. Competing approaches tend to solve this with brute force: oversized radiators, or spreading a light computing load across a massive constellation of satellites. Both add cost, mass, and complexity.
When you don't fight physics, it's cheaper, it weighs less, and it's more efficient.
Rob DeMillo, CEO, Sophia Space
Sophia's answer is its TILE platform, which distributes heat and computing power across each tile so no single point overloads. Instead of bolting on ever-larger radiators or diluting compute across hundreds of spacecraft, the architecture is designed to manage thermal load intrinsically. The claimed result is hardware that is cheaper, lighter, and more efficient than conventional orbital compute — a design philosophy the company sums up as working with the physics of space rather than against it.
A Third Path in the Orbital Compute Race
Sophia is not alone in eyeing orbit as the next frontier for computing. SpaceX has signaled its own ambitions for space-based data centers, and a growing field of startups and established players is exploring the category. But DeMillo argues the giants will build primarily for themselves — using orbital compute for internal workloads or a handful of anchor tenants — leaving a gap for open, generic infrastructure that anyone can use.
That is the position Sophia is staking out: not the brute-force hyperscaler model, and not the thin-compute-across-a-mega-constellation model, but a third path of efficient, generic data centers offering colocation and managed hosting in orbit. It is the space equivalent of a neutral cloud provider — selling capacity to Earth-observation operators, defense customers, and anyone else who needs to process data where it is created.
| Approach | How It Handles Compute & Heat | Tradeoff |
|---|---|---|
| Brute-force hardware | Oversized radiators on powerful nodes | Heavy, costly, hard to scale |
| Mega-constellation | Thin compute spread across many satellites | Coordination overhead, diluted capability |
| Sophia's TILE | Heat and compute distributed within each tile | Unproven at scale, must flight-qualify |
Why the Two-Flight Roadmap Is Smart
Sophia's sequencing reflects capital discipline. Software risk and hardware risk are separated: the fall 2026 SOOS flight on Kepler's network proves the operating system cheaply, without waiting to build and qualify custom silicon, while the 2027 TILE demo on an Apex Nova bus tackles the harder hardware and thermal validation once the software is de-risked. Leaning on Kepler for connectivity and Apex for the satellite bus lets a small, well-funded team reach orbit quickly without building an entire spacecraft from scratch.
What It Signals for the Space Economy
Sophia's raise is modest in dollar terms but outsized in what it represents. Compute is becoming an infrastructure layer of space itself — as fundamental as launch, buses, and communications. As sensors multiply and AI workloads move closer to the data, the ability to process information on orbit will help determine which missions are even possible. For founders building the strategic infrastructure of the next space economy, orbital compute is emerging as one of its most important and least-built layers.
The Bottom Line
Sophia Space's $7 million round, bringing total funding to $22 million, funds a disciplined two-step path to orbital data centers: a SOOS software flight on Kepler's network in fall 2026 and a TILE hardware demo on an Apex Nova bus in 2027. Its wager is that efficient thermal design, not brute force, unlocks affordable compute in space — and that an open, generic orbital data center is a foundational piece of the coming space economy.
Frequently Asked Questions
How much did Sophia Space raise and who invested?
Sophia Space is finalizing a $7 million SAFE financing round, which brings its total funding to date to $22 million. The round is backed by EverGreen — the NVIDIA Alumni Investment Network — along with SparkLabs Group and additional undisclosed investors. The company plans to use the capital for research and development, expanding its engineering and sales teams, and building a manufacturing pipeline to scale production.
What are TILE and SOOS?
TILE is Sophia Space's compute hardware platform for orbit, engineered to distribute heat and computing power across each tile so that thermal load is managed intrinsically rather than with oversized radiators. SOOS, the Sophia Orbital Operating System, is the software layer that runs and manages compute workloads on that hardware. Together they are designed to make orbital compute something customers can rent like a cloud service.
What is the significance of the Apex partnership?
Sophia has selected Apex to supply a Nova satellite bus for its 2027 demonstration flight, which will carry the TILE hardware and let it interact with onboard sensors to run AI inference workloads such as image processing in orbit. Using a commercially available bus lets Sophia focus on its compute payload rather than building an entire spacecraft, accelerating its path to a flight demonstration.
Why put a data center in space at all?
Satellites collect enormous volumes of data — terabytes to petabytes — but discard most of it because they lack onboard processing and cannot downlink it all through limited, expensive bandwidth. Processing data on orbit lets a spacecraft send a small, decision-grade answer instead of raw data, turning collected information into usable insight. As sensors and AI workloads proliferate, on-orbit compute becomes increasingly valuable.
How is Sophia different from SpaceX and other orbital compute efforts?
SpaceX and others are pursuing space-based data centers largely for their own internal use or a few anchor tenants. Sophia aims to offer open, generic data centers with colocation and managed hosting that any customer can use. Technically, it avoids the brute-force approach of oversized radiators and the alternative of thinly spreading compute across a mega-constellation, instead distributing heat and compute within each TILE for a design it argues is cheaper, lighter, and more efficient.