Inside The $100M Bet on the Future of Space | Northwood CEO on a16z
Summary
Northwood’s core bet is that ground infrastructure has become space’s “longest pole in the tent.” Satellites can now be built and launched faster than operators can connect to them; without command and data links, a spacecraft is “just like a rock in space.” Fixing the bottleneck requires integrating antenna R&D, sites, networking, software, and operations end to end.
Vertical integration compresses ground-station deployment from roughly three years to three months while turning bespoke capital equipment into shared infrastructure. Northwood designs antennas for standard shipping containers, commercial flights, bare dirt, and standard 240-volt power. Its platform model spreads one infrastructure investment across commercial, government, and allied missions instead of charging each customer for a one-off build.
The $50 million Space Force contract illustrates a shift toward commercial partners for proliferated systems. Northwood has been entrusted early with a significant program to modernize the Satellite Control Network, through which every U.S. government launch runs and missions spanning GPS, NASA, and missile tracking are supported. Mendler argues that venture can absorb some risk so government and commercial customers can pursue larger systems faster.
More satellites do not automatically produce more value: spacecraft are “really expensive depreciating assets,” and their ROI depends directly on ground throughput. With roughly 13,000 active satellites—an estimate Mendler said “sounds about right”—some government missions have launched without a ground plan, while commercial operators report that insufficient ground footprint limits how many customers they can serve.
Mendler considers optical inter-satellite links a “0% threat” because lower latency and friction should expand total space-data volume. The resilience challenge includes physical-site vulnerability, including in a Taiwan scenario; Northwood’s answer is proliferation—cheaper, rapidly deployed, volume-manufactured stations with enough regional redundancy that losing one site is not catastrophic.
The upside case is an enabling layer like the early internet, not a forecast of a particular application. Mendler says early internet builders could not foresee today’s internet but could build protocols and principles that unlock innovation. Northwood wants to support deeper-space missions, orbital compute, and data throughput potentially rivaling or surpassing internet volume. Execution rests on about 75 employees and a culture built to “accomplish unreasonable things on unreasonable timelines.”
Deep dive
1. Curiosity led Mendler from entertainment to space’s hidden bottleneck
Mendler’s career logic is deliberately short-horizon: follow curiosity, pursue it “to the nth degree,” and look only as far ahead as the horizon permits. Planning an entire 20-year career, she argues, constrains the possibilities; while acting, she could see perhaps five years forward and certainly did not expect to run a space company.
Entertainment gave her a platform for causes, but she wanted her impact to “go deeper.” Having watched the internet and cellular networks transform the world, she saw space networking as another potentially fundamental layer. She and her husband were both ambitious and curious about how the world works, and both were passionate about building businesses, so they aligned their time and energy on Northwood.
Northwood began with antenna experiments and a Home Depot run during the pandemic. Receiving signals hundreds of miles away—“fainter than the power of a flashlight”—made the technology feel magical; a subsequent study of commercial and government operators exposed an approaching ground-capacity bottleneck.
2. Owning the whole ground stack turns three years into three months
Mendler’s framing: faster launch cadence and spacecraft manufacturing exposed the neglected third pillar. Antenna makers supplied point solutions, while software integrators depended on infrastructure laid by others; no stakeholder had incentives to optimize the complete ground system. Northwood concluded that “the only way to address fixing the ground segment was to do the whole thing.”
Timing also changed as missions became more proliferated, transmitted more data, and moved dynamically in and out of different orbits. Older, more static missions could better tolerate a ground segment that did not keep pace; newer commercial and government missions require a new ground architecture.
That whole includes antenna R&D, RF conversion into usable data, land procurement and site development, networking, command software, and global operations. It is a “massive undertaking” requiring capital, multidisciplinary talent, and orchestration—but aligning Northwood’s success with the customer’s mission success requires owning every dependency.
Erik’s deployment question surfaces the practical advantage. Many traditional ground systems involve bespoke equipment, supply-chain delays, ocean shipping, permits, concrete foundations, and multi-story construction; Northwood instead specifies a standard shipping container carried on “a commercial United Airlines flight,” deployment on “a patch of dirt,” standard 240-volt power, and system-wide telemetry for startup within minutes.
Erik’s SpaceX comparison—why has ground not achieved launch’s order-of-magnitude cost decline?—gets the same answer: vertical integration enables a standardized system serving commercial, government, and allied missions. As a shared platform, one Northwood investment supports many customers, replacing large one-off capital expenditures with smoother costs and accumulated learnings.
3. More links and more sites strengthen the ground thesis
Asked whether Starlink’s direct optical inter-satellite links threaten ground stations, Mendler answers “0% threat.” Her directional bet is growing space-data volume: inter-satellite links lower latency and transmission friction, opening applications such as internet service that can sometimes match or beat internet latency speeds.
Erik’s Taiwan-scenario pushback—ground stations can become targets—makes resilience a commercial and government concern. Northwood follows the proliferation model used by teammates from Starlink: make stations cheaper, faster, and manufacturable at volume, then place multiple sites in a region so one loss does not take the service offline.
The early-internet analogy holds because builders can bet on a direction without knowing the eventual applications. Mendler points to TCP/IP protocols and designing layers to support innovation. Space’s enabling layers include launch, power, propulsion for maneuvering, and connectivity back to Earth.
4. Government demand and orbital ambition pull the same infrastructure forward
On orbital data centers, Mendler emphasizes enablement rather than prediction. The relevant details include whether the use case is training or inference, alongside other unresolved questions, and she says valid concerns may mean long timelines. Northwood wants to understand ambitious missions well enough to help “translate them from dream to reality as fast as possible.”
She maps future opportunity along two axes. One is altitude—stretching the tether beyond low Earth and geostationary orbit toward deeper exploration; the other is throughput—asking what becomes possible if space data could match or surpass internet volume. AI might help unlock the “massive treasure trove of data about our planet” that remains underused.
Public and private development remain analogous to the internet era, but venture now absorbs more risk at the frontier. The $50 million Space Force award illustrates that shift: procurement models built for slower, bespoke systems cannot meet the urgent increase in proliferated ground capacity, so government is looking for commercial partners to absorb some of that risk.
The program targets the Satellite Control Network, a common resource supporting every U.S. government launch and missions including GPS, NASA, and missile tracking. Economically, the logic is blunt: a satellite begins depreciating at launch, its data creates the value, and data output is “directly proportional” to ground connectivity—making ground quite literally a lever on spacecraft ROI.
5. Global scaling demands multidisciplinary talent and unusual ownership
Northwood is expanding beyond the phased arrays for which it is known, with additional ground products in development alongside a global network. At recording, it had five international entities, operated on two continents, and expected to reach several more before year-end—Mendler jokes it will become “the United Nations of Northwood.”
The roughly 75-person company had doubled several times and hoped to double again during the year. Its hiring spans global site builders from Starlink and analogous deployment systems such as AT&T cell towers and Tesla Superchargers, plus supply-chain specialists and a substantial software organization covering networking, embedded systems, and the front end.
The first cultural expectation is to “accomplish unreasonable things on unreasonable timelines,” but Mendler distinguishes that from brute force. Speed comes from understanding a problem deeply enough to identify acceptable trades, take smart risks, and apply “not just force” but “also cleverness.”
End-to-end ownership means caring beyond a job description—as when teammates repeatedly stayed awake more than 24 hours during a week deploying Northwood’s first antenna in North Dakota. The final standard is “a categorical outcome, not just an incremental outcome,” supported by low ego, trust, admitting faults, and enough boldness to raise flags when problems arise.