# Chapter 5 — The data gravity problem > Eventually, moving all raw planetary data to Earth becomes irrational. - Source: https://planetary.ravisuhag.com/the-data-gravity-problem - Published: 2026-05-27 - Author: Ravi Suhag - Part of: Planetary Intelligence (How Earth observation, AI, and orbital compute converge.) --- > *Eventually, moving all raw planetary data to Earth becomes irrational.* For most of the history of Earth observation, satellites were treated as cameras. Capture first, transmit later, process on Earth. The spacecraft existed primarily as a sensing device — intelligence lived on the ground. That architecture made sense when observation volumes were relatively small. But the economics begin collapsing once the planet becomes continuously observable across multiple sensing modalities — high-frequency imaging, hyperspectral sensing, orbital video, persistent SAR coverage, thermal monitoring, multi-constellation fusion. The amount of raw planetary data being generated is growing faster than the infrastructure designed to move it. Eventually, moving all raw planetary data to Earth becomes irrational. Not difficult. Irrational. Because transmission itself becomes the bottleneck. the cost curve flips. moving the data costs more than computing on it where it was born. ## The hidden constraint of Earth observation Most people outside the industry think the hard part of EO is launching satellites. It is not. Launch costs are falling, satellite manufacturing is becoming standardized, and sensor technology continues improving. The real constraint is increasingly invisible. Bandwidth. the constraint nobody markets. it doesn't appear in datasheets, but it bounds everything downstream. A satellite can only transmit data to Earth when it has line-of-sight access to a ground station or relay network. Even then, transmission capacity remains fundamentally limited by physics, spectrum allocation, power budgets, antenna design, and orbital geometry. Meanwhile, sensor generation rates are exploding. Modern sensing systems can generate far more data than can realistically be transmitted continuously to Earth. This creates a structural asymmetry. Observation scales faster than downlink capacity. Eventually, every EO company encounters the same realization: you cannot transmit everything. So the question changes. Not "how do we collect more data" but "what data is actually worth transmitting". That is the beginning of orbital computation. ## Downlink bottlenecks Historically, downlink systems were treated as infrastructure problems. Build more ground stations, lease more bandwidth, improve compression pipelines. But the scale transition underway changes the nature of the problem entirely. Imagine persistent global video coverage from orbit, or continuous hyperspectral monitoring across agricultural regions, or large-scale SAR constellations operating around the clock. The raw data volumes become staggering — petabytes per day, then exabytes. At that scale, transmitting raw sensor output becomes economically and operationally unsustainable. Not because storage is expensive — because movement is expensive. Every transmitted bit consumes power, spectrum, orbital scheduling capacity, antenna time, and infrastructure coordination. This is the same realization hyperscale cloud systems encountered years ago. Data movement eventually dominates system architecture. And once movement dominates, compute migrates toward where the data originates. ## Space-to-ground bandwidth constraints Earth-to-orbit communication still behaves more like early internet infrastructure than modern cloud networking — intermittent connectivity, limited throughput, high latency windows, strict scheduling constraints. A satellite does not maintain persistent broadband connectivity with Earth. It passes over communication windows periodically. This means the spacecraft itself increasingly becomes a buffering system. Data accumulates onboard faster than it can always be transmitted. As sensing density increases, this imbalance worsens — more sensors create more observations, more observations create more backlog, more backlog increases operational filtering pressure. Eventually, the spacecraft must decide what should be prioritized, what should be compressed, what should be discarded, what requires immediate transmission. These are no longer transmission decisions. They are intelligence decisions. the moment the satellite chooses what to send, it is already reasoning. the camera has become an editor. ## Storage and transmission economics One of the most important shifts in computing history occurred when engineers realized storage was becoming cheaper faster than bandwidth. Earth observation is entering a similar transition. Orbital compute hardware is improving rapidly, onboard storage density continues increasing, and edge AI accelerators are becoming viable in space-qualified systems. Meanwhile, transmitting massive raw datasets to Earth remains constrained. This changes the economics fundamentally. It may soon become cheaper to process data in orbit than to transport it blindly to Earth. This is a profound architectural inversion. The satellite stops behaving like a remote camera. It starts behaving like a distributed compute node. This is the real significance of onboard AI — not simply faster inference, but selective cognition. The system learns what matters before transmission occurs. inference at the sensor. the bit that never gets sent is the cheapest bit in the system. ## Latency challenges Transmission delays are not merely bandwidth problems. They are operational problems. Many future EO applications require near-real-time responsiveness — disaster monitoring, military awareness, maritime intelligence, infrastructure anomaly detection, climate event tracking, autonomous mission coordination. If sensing systems must wait for full downlink cycles before processing occurs, decision latency increases dramatically. In some operational contexts, the delay itself destroys the value of the information. A wildfire spreading rapidly does not care about orbital scheduling windows. A military maneuver does not pause until imagery reaches a cloud region. A collision risk in orbit cannot wait for centralized analysis pipelines. This forces computation closer to observation. The same way autonomous vehicles cannot depend entirely on remote cloud reasoning, future orbital systems cannot rely exclusively on Earth-based compute loops. Local reasoning becomes necessary — not because it is elegant, but because physics demands it. ## The distributed systems parallel The future architecture of Earth observation looks increasingly similar to distributed cloud computing. This is not accidental — the same forces are emerging again under different physical constraints. Large-scale cloud systems evolved because centralized architectures stopped scaling efficiently. Moving all computation through a single location became too slow, expensive, and operationally fragile. So compute distributed outward — closer to users, closer to applications, closer to data generation itself. Edge computing emerged from this reality. Earth observation is now replaying the same evolution at planetary scale. Satellites become edge nodes. Ground stations become regional ingress points. Orbital relays become network infrastructure. Constellations become distributed sensing clusters. The planet itself starts behaving like a continuously updating distributed system. Eventually, entirely new architectural questions emerge. How do orbital systems coordinate state? How do satellites share learned representations? How do constellations synchronize models efficiently? How does inference occur collaboratively across fleets? These are not traditional aerospace problems. They are distributed systems problems appearing in orbit. ## From spacecraft to orbital infrastructure This transition changes the identity of satellites themselves. Historically, spacecraft were specialized hardware assets — carefully engineered, individually operated, mission specific. But once onboard compute, continuous sensing, and distributed coordination become central, satellites begin behaving more like programmable infrastructure. Less like standalone machines. More like nodes in a planetary compute network. This mirrors the evolution of servers. A single server once mattered deeply. Today, infrastructure is defined by orchestration across massive distributed fleets. The same abstraction shift is beginning in orbit. The future operator may not think in terms of satellites at all. They may think in terms of observation capacity, inference coverage, temporal resolution, planetary state synchronization, and distributed reasoning throughput. The infrastructure layer becomes computational rather than purely aerospace. the unit of work stops being "a satellite". it becomes a slice of compute that happens to be in orbit. ## Intelligence before transmission The important realization is that raw data is not the end product. Intelligence is. And intelligence does not require transmitting every photon collected from orbit — only the meaningful abstractions. A future EO system may never send most of its raw observations to Earth at all. Instead, it may transmit detected anomalies, infrastructure state changes, behavioral predictions, environmental alerts, object trajectories, operational summaries, learned embeddings, and compressed world representations. The output becomes semantic rather than visual. This is the same transition that transformed modern computing systems everywhere else — raw signals become structured understanding. Eventually, Earth observation systems stop functioning primarily as imaging pipelines. They become distributed planetary cognition systems operating partially in orbit itself. And at that point, the distinction between space infrastructure and compute infrastructure starts disappearing entirely.