Planet Labs: Change of Narrative
Recent Developments at Planet Labs - From Space Operator to AI Infrastructure Play
Planet Labs, which featured in my June 2025 post Small Satellites Here, There, Everywhere, has undergone a dramatic transformation in the second part of the year — from a struggling satellite imagery provider into what investors now view as an AI infrastructure play. Following the company's FY26-Q3 (ended October 31) earnings announcement on December 10, the stock surged approximately 35%, reaching new 52-week highs above $19 per share ($6b cap) and trading at an eye-popping 20x forward sales.
The Catalyst: Explosive Backlog Growth
The primary driver behind this revaluation was a 361% year-over-year increase in Remaining Performance Obligations (RPO), which reached $672m, while total backlog surged 216% to $735m. This backlog now represents nearly 3x Planet's annual revenue, providing unprecedented visibility into future revenue streams. Approximately 61% is expected to convert to revenue within the next 2 years, fundamentally altering the company's growth trajectory.
The Q3 results themselves were impressive:
Revenue: Record $81.3m (+33% YoY), crushing analyst estimates of $72.2m
Defense sector growth: Over 70% YoY — a major driver
Free cash flow: $0.9m (third consecutive positive quarter)
Capital raise: $460m in 0.50% convertible senior notes
The Thematic Shift: Data Monopoly Meets Geospatial AI
What fundamentally changed investor perception was the recognition that Planet possesses something irreplaceable: a temporal moat built on over a decade of daily global Earth scans. With an average of 1,700 images of every place on Earth in its archive, Planet has created a historical dataset that new competitors cannot replicate.
Hyperscalers, particularly Google, are increasingly viewing this daily Earth observation data as a finite but essential resource for training geospatial AI models. The market is no longer valuing Planet for its satellites or imaging hardware. It is pricing in the scarcity and uniqueness of Planet's proprietary dataset for predictive AI modeling.
Strategic Partnerships: Google's Project Suncatcher
In November, Planet announced a groundbreaking partnership with Google for Project Suncatcher, Google's ambitious research initiative to deploy space-based AI compute infrastructure. Planet will build and operate two prototype satellites equipped with Google's Tensor Processing Units (TPUs), targeting launch by early 2027. These satellites will test the viability of running AI workloads in orbit, powered by solar energy and connected via high-bandwidth free-space optical links capable of 1.6 terabits per second.
The Case For Space Data Centers
Space data centers sound like science fiction. If I were a betting man, I would take the under. However, respected tech experts such as Gavin Baker (Atreides Management) and Sundar Pichai (Google) have vocally supported the idea. In the December 9 episode of Invest Like the Best, Baker made a compelling, first-principles argument for why space data centers are not only viable but inevitable and superior to Earth-based facilities. He went as far as saying that space data centers will be:
the most important thing that's going to happen in the world in the next 3 to 4 years
The three pillars of his argument are:
Unlimited solar power: Satellites can remain in sunlight 24/7 with 30% more intense solar irradiance (no atmospheric filtering)
Free cooling: Simply place radiators facing away from the Sun
Lower latency: Free-space optical lasers travel through vacuum faster than light through fiber
Baker draws on deep technical expertise (semiconductor analyst background) and direct investment experience (led Fidelity's investments in SpaceX, was largest shareholder of Nvidia at $2b market cap, and largest shareholder of Tesla when it was a struggling startup).
Now, I like Gavin Baker and don't doubt his knowledge of technology and data centers. However, he has horses in the race, which may be coloring his judgment. There are some substantial and well-known criticisms being raised against 2 of his 3 pillars.
On cooling: Jen Zhu (co-founder of Power Dynamics, an AI infrastructure advisory firm) delivered one of the sharpest technical rebuttals via this tweet. Her core point:
Space is cold ≠ cooling in space is easy
In space's vacuum, heat can only dissipate through thermal radiation, which is slow and inefficient compared to Earth-based conduction and convection. She estimates that for a 1 GW data center in space, you would need radiators kilometers wide/long — an enormous amount of material to launch and assemble.
On latency: While Baker is correct that vacuum propagation is 1/3 faster than fiber, critics argue this advantage becomes irrelevant when accounting for network topology. Modern AI training requires massively parallel, all-to-all communication across thousands of compute nodes. On Earth, dense network topologies provide rack-to-rack communication in ~7-10 microseconds. In space, thermal constraints mean each satellite can only support one rack's worth of compute, forcing satellites to be separated by hundreds of meters to kilometers. With limited inter-satellite links (4-6 per satellite), reaching distant nodes requires 10-20+ hops, each adding 5-10 microseconds of switching latency. The sum total result is up to 20x worse than terrestrial data centers despite faster propagation.
Interestingly, Google's own technical paper on Project Suncatcher directly acknowledges these challenges:
High-bandwidth inter-satellite links require our satellites to fly in a much more compact formation than any current system. Achieving data center-scale bandwidth requires links between satellites that support tens of terabits per second... However, achieving this kind of bandwidth requires received power levels thousands of times higher than typical... we can overcome this challenge by flying the satellites in a very close formation (kilometers or less).
My Take: These are hard problems — not necessarily unsolvable, but definitely not trivial. It will be fascinating to watch developments in this space. The partnership positions Planet at the intersection of two transformative trends — space infrastructure and edge AI computing. It is a nice corner from which to enjoy the benefits of future AI infrastructure spend. However, I try not to get too starry-eyed about what remains largely unproven technology.
AI Acceleration: Bedrock Research Acquisition
In November, Planet acquired Bedrock Research, a Denver-based geospatial AI solutions company specializing in remote sensing, AI/ML, and national security applications. Bedrock's team brings proven expertise in deploying deep learning algorithms for real-time analytics, with their CEO having previously led multi-hundred-million-dollar programs for the US intelligence community.
The acquisition followed a year of close partnership during which Bedrock demonstrated highly efficient AI workflows for applying advanced machine learning to Planet's data. This capability is particularly valuable for Planet's expanding Defense & Intelligence business, enabling automated object detection, predictive analytics, and real-time threat identification. The integration allows Planet to accelerate development of AI-driven global monitoring solutions generating strong demand from government customers.
Fleet Expansion and AI-Enabled Satellites
Planet has been aggressively expanding and upgrading its satellite constellation with AI capabilities integrated directly at the edge.
Pelican High-Resolution Fleet: Planet successfully launched multiple Pelican satellites throughout 2024-2025, including Pelican-2 (January 2025), Pelican-3 and Pelican-4 (August 2025), and Pelican-5 and Pelican-6 (November 2025). These satellites provide up to 40 cm resolution imagery across six multispectral bands and are equipped with NVIDIA Jetson AI platforms for on-orbit edge computing. The goal is to reduce the time between data capture and delivery from hours to minutes — critical for applications ranging from security to disaster response.
In January 2025, Planet secured a massive $230 million multi-year contract with SKY Perfect JSAT (Asia's largest geostationary satellite operator) for Pelican imagery. This marked Planet's entry into the satellite-as-a-service business model, where JSAT funds the satellite construction but Planet retains global data resale rights. It is essentially getting customers to finance constellation expansion while maintaining data monetization.
Owl Next-Generation Monitoring Constellation: In October, Planet unveiled its most advanced satellite program yet. This next-generation fleet will provide near-daily, 1-meter class resolution imagery — a significant improvement over the 3-meter resolution of the current SuperDove constellation. Each Owl satellite will be equipped with NVIDIA GPUs to perform AI processing directly in orbit, enabling onboard analysis such as object identification, anomaly detection, and activity flagging at the sensor's edge. This dramatically reduces latency, with the goal of delivering insights to customers within an hour rather than days.
Tanager-1 Hyperspectral Mission: Planet continues operating Tanager-1, launched in August 2024 as part of the Carbon Mapper Coalition. This satellite combines Planet's smallsat bus technology with NASA JPL's state-of-the-art imaging spectrometer to detect and quantify methane and CO₂ super-emitters globally. By November 2024, Carbon Mapper had released over 300 initial methane and carbon dioxide plume detections from Tanager-1, demonstrating unprecedented capabilities for emissions monitoring across oil and gas, coal, waste, and agriculture sectors.
Defense and Government Contract Momentum
Planet's Defense & Intelligence business has emerged as a major growth driver, with government revenue increasing 70% YoY in FY26-Q3. Major contract wins include:
€240m Germany contract: PlanetScope data, AI-enabled solutions, and dedicated Pelican satellite capacity over Europe
$230m SKY Perfect JSAT (Japan): Customer-funded Pelican satellite construction with Planet retaining global data resale rights
$12.8m NGA Luno B contract: AI-enabled vessel detection and monitoring (AAMOR) across Asia-Pacific
$13.5m NASA task order: PlanetScope data under Commercial Satellite Data Acquisition contract
$13.2m NRO contract renewal: Baseline PlanetScope data with framework for high-resolution Pelican imagery
$7.5m U.S. Navy renewal: Vessel detection and monitoring
Seven-figure NATO expansion: Persistent space-based surveillance and Maritime Domain Awareness
These wins reflect a broader trend toward sovereign geospatial intelligence. Nations are seeking to avoid over-reliance on foreign satellite data following the Ukraine conflict and shifting geopolitical dynamics. Multiple European countries are designing their own high-resolution satellite constellations (Germany's €1.7b SPOCK-1 SAR system, ESA's proposed €1b Earth Resilience from Space program, Italy's €1.1b IRIDE constellation, and the Kongsberg-Helsing intelligence constellation), but these won't be operational until 2028-2030. Planet's contracts appear to be bridge solutions, providing immediate sovereign capacity while Europe builds out its own infrastructure.
Looking Ahead
Planet Labs has successfully repositioned itself at the nexus of three secular trends:
The global race for sovereign geospatial control
AI-driven monetization of Earth observation data
Space-based edge computing infrastructure
Execution will be critical. Planet must successfully convert its record backlog into revenue, deploy its next-generation Owl constellation on schedule, demonstrate the value of AI-powered analytics through the Bedrock integration, and continue winning large government contracts in an increasingly competitive market. The Project Suncatcher partnership with Google could open entirely new revenue streams if space-based AI compute proves viable at scale.
For investors with a multi-year horizon willing to embrace volatility, Planet offers exposure to structural trends that appear durable regardless of short-term economic cycles. For value investors seeking margin of safety, however, there is nothing to see here — just expensive metal in space trading at AI multiples. But don’t worry! The value program will continue soon. Until then 🔔🎁🎄☃️
May your margin of safety be generous,
your downside protected,
and your upside pleasantly mispriced.
Have a wonderful holiday season!