← back to ranking

PL · Planet Labs PBC

Aerospace & Defense · mkt cap $15.5B · calls: Q1 FY2027 vs Q4 FY2026
62.0 conviction · conf-adj 54

conf 4/10 🚀 reported partial

enthusiasm:24.0 · trend:-5 · quantifies:12 · impact:8 · under_radar:0 · credibility:12 · business_impact:8 · disruption:0 · commitment:0 · confirmation:3

Enthusiasm latest 8 / prev 9 (falling)

Planet Labs frames its AI thesis around two compounding layers: near-term AI-enabled analytics (Maritime Domain Awareness, Global Monitoring Service, the new natural-language AI app, SuperRes) that are already driving D&I revenue growth above 50-65% YoY and beginning to re-accelerate commercial bookings, and a longer-horizon bet that its daily-scan archive is 'foundational to real-world models' the way Wikipedia was to LLMs, positioning Planet as an AI data utility. The thesis is partially credible — Bedrock integration, NVIDIA GPU speedups, and specific contract awards (NGA $21.9M for advanced analytics) show genuine AI revenue, and SuperRes is a shipped product — but management has not disaggregated AI-attributable revenue from total D&I/commercial growth, leaving the monetization narrative more visionary than verified. Enthusiasm stepped down marginally from Q4 FY2026's evangelical 'first space-and-AI company' framing to Q1 FY2027's more product-delivery-focused tone, suggesting the transition from AI story to AI execution is underway but not yet complete.

GROUNDED NEXT-FY IMPACT vs CONSENSUS

Grounded on actual base — revenue $0.3B · net income $-0.2B · net margin -80.2% · diluted EPS -0.8

These are next-fiscal-year annual uplift estimates, not next-quarter numbers.

Aggregate next-FY est. rev uplift: 7.12% · next-FY EPS uplift: % · vs analysts: unclear · priced in: high (model's call-read: medium; verdict above is the hard-data one used for ranking) · confidence: 4/10

ClaimFigureArithmeticNext-FY Rev %Next-FY EPS %
Bedrock: 600 sites in 3h vs weeks
productivity · soft		600 sites in 3 hours vs weeksDeployment-speed gain; no $ cost base or revenue figure disclosed to monetize. Faster standup could add capacity but unquantified -> no arithmetic possible.
NVIDIA GPU ~100x ground pipeline speedup
productivity · soft		~100x on certain pipeline componentsCompute-time/throughput speedup on parts of pipeline; no opex/COGS $ or headcount base disclosed to convert into a saving. 'Faster answers to clients' qualitative -> unanchored.
SuperRes: PlanetScope 3m -> 2m class
engagement · soft		3m -> 2-meter class resolutionAI product/resolution enhancement that may support upsell/ASP, but no attach rate, price uplift, or unit base disclosed -> cannot size revenue without invented elasticity.
Pelican Gen2 onboard AI: hours -> minutes latency
other · soft		hours to minutes (tech demo)Onboard-AI latency improvement, demo-stage, no associated contract/price/cost figure -> unanchored.
Owl: 1m daily scan, ~10x data, ~10x latency
other · soft		1m vs 3m; ~10x data; ~10x latency (forward demo)Forward tech-demo capacity/latency multipliers, not $ revenue; no next-FY bookings/ARPU base to apply against -> cannot translate to next-FY revenue.
NGA Luno B maritime AI-analytics award
revenue		$21.9M, 1-yearrev_uplift_pct = 100 * 21,900,000 / 307,727,000 = 7.12%. 1-year award (Q1 FY2027) recognized ratably ~$21.9M over the next FY -> full amount lands in next FY (revenue, not multi-year booking). Adopter-side. GAAP NI = -$246.9M (loss-making) so EPS% off it is meaningless -> null.7.12
Nontechnical users build bespoke app in <1 hour
engagement · soft		bespoke app in under an hour (forward)Forward TAM-expansion / barrier-to-entry claim; no users, conversion, or ARPU figure -> unanchored.

Assumptions: Next FY = fiscal year ending 2027-01-31. NGA $21.9M: 12-month performance obligation recognized ratably (~$21.9M/yr) -> entire $21.9M attributed to next FY as revenue, not a one-time booking spike. Only one claim is anchored. EPS: GAAP NI = -$246.9M (net margin -80.2%), so any EPS-uplift % off it is an artifact -> est_eps_uplift_pct null per loss-making guardrail (illustrative: at 35% incremental analytics/services margin NGA adds ~$7.67M NI, ~38% of the -$19.9M FY2026 adjusted consensus loss, but denominator too distorted to report). Tax 21% irrelevant (loss-making). All other AI claims are capability/speed/product-quality statements with no disclosed $ base, price, volume, or headcount -> soft=true, not invented. No supplier-side AI compute revenue quantified.

Top line: Only one AI claim is monetizable today: the $21.9M NGA Luno B maritime-analytics award = 7.12% of FY2026 revenue ($21.9M/$307.73M; 7.34% vs $298.32M consensus avg), and only if treated as fully ratable into next FY and incremental to run-rate. The rest of the AI story (SuperRes, Pelican/Owl onboard compute, Bedrock/NVIDIA speedups, <1hr app builder) is product-quality and capability signaling with no disclosed price, attach rate, or volume — real strategic optionality but $0 quantifiable next-FY revenue. Headline adopter uplift ~7%, resting entirely on a single contract. No supplier-side AI compute sales quantified.

Bottom line: EPS impact is not computable: GAAP net income is -$246.9M (net margin -80.2%), so any EPS-uplift % is an artifact of a deeply negative base -> set to null. At ~35% incremental analytics margin the NGA award adds ~$7.67M net income, which narrows but does not close the loss. Management frames the FY as 'the start of returns on AI investments,' i.e. AI is still net-cost. Productivity claims (100x compute, 600 sites in 3h, Owl latency) plausibly trim cloud/ops cost but no opex base is disclosed to size them -> 0% modeled.

[EPS uplift n/m (loss-making base)] Consensus already implies strong growth: revenue $235.8M (FY24) -> $244.8M (FY25) -> $298.3M (FY26) consensus vs $307.7M actual (beat by ~$9.4M, +3.15%); EPS path improving (-0.47 -> -0.18 -> -0.079). The one hard AI item (NGA $21.9M = 7.12% of revenue) is a publicly announced Q1 FY2027 award likely already in defense/backlog forward models, so it does not clearly sit above consensus. No FY2027 revenue/EPS consensus in the table to prove AI above forward estimates. With ~7% of the revenue base anchored and the rest unmonetized vision, the AI thesis is roughly in line with what's priced.

MODEL CONSENSUS (impact)

partial

Strong agreement: single $21.9M NGA anchor (~7.1% rev), all else soft, EPS null on loss-making base. Minor rounding and claim-type differences reconciled toward the better-justified, more conservative labels.

Conflicts reconciled
FieldOpus 4.8GPT-5.5
Rev uplift %7.1–
EPS uplift %––
Priced inmedium–
vs analystsunclear–
Confidence4–
Top lineOnly one AI claim is monetizable today: the $21.9M NGA Luno B maritime-analytics award = 7.1% of FY2026 revenue ($21.9M/$307.7M), and only if treated as fully ratable into next FY. The rest of the AI story (SuperRes, Pelican/Owl onboard compute, Bedrock/NVIDIA speedups, <1hr app builder) is product-quality and capability signaling with no disclosed price, attach rate, or volume — real strategic optionality but $0 quantifiable next-FY revenue. Headline adopter uplift therefore ~7%, resting entirely on a single contract.–
Bottom lineEPS impact is not computable: GAAP net income is -$246.9M (net margin -80.2%), so any EPS-uplift % is an artifact of a deeply negative base — set to null. Management explicitly frames FY as 'the start of returns on AI investments,' i.e. AI is still net-cost. The $21.9M contract at even high software margin (~56% gross) would only modestly narrow a ~$247M loss; the productivity claims (100x compute, 600 sites in 3h) plausibly trim cloud/ops cost but no opex base is disclosed to size them.–
ReasoningConsensus already implies strong growth: FY25 $244.8M -> FY26 $298.3M avg = +21.9%, with 6 analysts covering. The one hard AI item (NGA $21.9M = 7.1% of revenue) is a publicly announced Q1 FY2027 award likely already in those forward numbers, so it does not clearly sit above consensus. EPS consensus is itself improving toward breakeven (-0.473 -> -0.177 -> -0.079), consistent with AI being a cost-reduction tailwind but not a quantified earnings driver. With ~6% of the revenue base anchored and the rest unmonetized vision, the AI thesis is roughly in line with what's priced.–

Rows highlighted where the two models disagreed.

QUANTIFICATIONS
Bedrock AI deployment speed: 600 monitoring sites in 3 hours vs. weeks previously (Q4 FY2026 (reported), bottomline)
“The team is helping us scale rapidly and deliver AI-based solutions, notably standing up 600 new monitoring sites within three hours compared to a weeks-long process when we first launched the service.”
NVIDIA GPU processing speedup on ground pipeline: ~100x on certain pipeline components (Q4 FY2026 (early results), bottomline)
“we have seen some early results that are very promising, with big speedups like 100x on certain parts of our codebase. Getting answers to our clients faster is really important.”
SuperRes resolution improvement: PlanetScope improved from 3m to 2-meter class via AI (Q1 FY2027 (launched), topline)
“We also launched a new feature called SuperRes, an AI-powered technology to improve the resolution of our plan scope data into a 2-meter class resolution visual product”
Pelican Gen 2 analysis latency reduction via onboard AI: Hours to minutes (Q1 FY2027 (tech demo phase), topline)
“the NVIDIA chips that we're flying with the other ones. In combination is what enables the more real-time insights going from hours, latency for getting analysis of after you take a picture to minutes.”
Owl constellation resolution improvement: 1-meter daily scan vs. current 3-meter; ~10x more data per unit area; ~10x latency improvement (Q4 FY2026 (forward guidance on Owl tech demo), topline)
“we are moving towards one-meter scan rather than three-meter, and that is roughly 10 times more data per unit area of the ground. And roughly a 10x improvement in latency as well because they will be equipped with both onboard compute systems as well as satellite-to-satellite comms.”
NGA AI-analytics contract value (AI-specific award): $21.9 million, 1-year (Q1 FY2027, topline)
“the U.S. National Geospatial Intelligence Agency, the NGA, awarded Planet a $21.9 million 1-year contract extension for maritime surveillance under the Luno B IDIQ for advanced analytics for maritime operations and connivance.”
Time-to-value for nontechnical AI app users (forward): Bespoke application buildable in under an hour (Q4 FY2026 (near-term expectation), topline)
“we think that more generic AI solutions will soon empower nontechnical users to go from a concept to a bespoke application in under an hour”
PAST (realized)
CURRENT (now)
FORWARD (guidance)
TRACK RECORD — PROMISE vs DELIVERY

76/100 track record   too-early  6 calls reviewed

Planet talks up AI constantly but rarely commits to hard dated KPIs; the one clear near-term product rollout (Tanager commercialization) was met, while newer quantified targets (Owl, SunCatcher, sub-hour app builder) are still pending or only in beta, so quantified AI credibility is largely unproven rather than disproven.

Deploy 2 SunCatcher prototype satellites with Google TPUs for scaled in-space AI compute by early 2027 — promised Q3 FY2026
too-early As of Q1 FY2027 the program remains in funded R&D with no launch update; early 2027 deadline has not yet arrived.
Launch first Owl tech demo with onboard NVIDIA GPUs and ~1-meter-class daily monitoring by later calendar 2026 — promised Q3 FY2026
too-early Through Q1 FY2027 Owl is still pre-demo while Gen 2 Pelican work advances; no Owl demo launch reported yet with calendar 2026 still open.
Enable nontechnical users to go from concept to a bespoke AI application in under 1 hour — promised Q4 FY2026
partial Q1 FY2027 launched a private-beta natural-language AI app for archive search and reports, but full sub-hour bespoke-app workflow for general users is not yet commercially available.
Stand up 600 new AI-enabled global monitoring sites within 3 hours (vs weeks-long setup) — promised Q4 FY2026
partial Management reported Bedrock integration achieving this operational metric in Q4 FY2026, but no prior call had set this as a forward quantified target to track against.
Offer Tanager hyperspectral methane/CO2 analytics commercially to the broader energy and civil-government market within the next few months — promised Q4 FY2025
delivered Subsequent calls show broader Tanager commercialization—California STPP methane program, Carbon Mapper scaling to ~3,000 plume sources, and expanding civil-government sales—consistent with the stated near-term rollout.
Fine-tune Anthropic Claude on Planet satellite archive to measurably enhance model accuracy on geospatial tasks — promised Q4 FY2025
quietly-dropped The Anthropic collaboration is not updated with accuracy metrics or milestones in later calls as NVIDIA/Google AI partnerships dominate the roadmap.
PRICED-IN (REFINED)
HIGH (already in)

Est. revisions rising  ·  Fwd P/E -76.8  ·  EV/Sales 36.5x

AI claim maps to Reportable Segment

Estimate momentum is rising: last-month price targets ($51.5) sit well above last-quarter ($40.25) and last-year ($23.57) averages, and forward consensus lifts FY26 revenue to ~$298M (+22% vs FY25) while losses narrow—so AI/growth upside is already being baked into numbers. Valuation is rich at ~36.5x EV/Sales and ~36x P/S on ~$12B market cap despite negative EPS (fwd P/E -76.8 is not meaningful but underscores loss-making status at a premium multiple). Planet reports a single product line (Reportable Segment), where AI-driven analytics and imagery monetization would flow. Rising revisions plus stretched multiples point to AI upside largely reflected in the price.
COVERAGE — ENTHUSIASM TRAJECTORY + CATALYSTS
6Q4 FY20257Q1 FY20268Q2 FY20268Q3 FY20268Q4 FY20268Q1 FY2027

AI enthusiasm across 6 calls — trend ↗ rising

AI moved from GMS R&D and partner MDA to AI-powered detection feeds and eight-figure defense analytics contracts.

RECENT AI CATALYSTS & NEWS
BUSINESS IMPACT - QUALITATIVE MATERIALITY

7/10 qualitative impact   material  medium-term · mixed evidence

Where AI matters: D&I analytics monetization & data-access layer

AI-enabled GMS/MDA, SuperRes, and defense awards (e.g., NGA $21.9M) are real shipped drivers of a fast-growing D&I line and commercial pipeline, but management does not break out AI-attributable revenue and the NL-query app remains early beta.

Caveats: No disaggregated AI revenue—growth narrative may overstate AI vs broader D&I/commercial mix; NL AI app still private beta; sub-hour bespoke-app claims unproven; Rich valuation and rising estimates may already embed AI upside; Competitors and customers can deploy similar GPU/CV stacks on rival or in-house EO data

AI DISRUPTION / CANNIBALIZATION RISK  tailwind · 3/10

Planet's moat is the proprietary daily global imagery archive and constellation cadence, which generic LLMs and open CV cannot replicate; AI mainly lowers delivery cost and widens access, with only modest risk that commoditized analytics compresses software/margin on the insight layer.

OPTIONS / MARKET STRUCTURE

option liquidity: good

ATM IV
—
TYPICAL BID-ASK
—
OPEN INTEREST
—

proxy inputs — dollar-ADV $490M · beta 1.99662 · px $36.28

source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.

CONFIRMATION — INSIDERS · 13F · LANGUAGE
Mixed — insiders selling, institutions adding, management language 5/10 measured.
INSIDERS selling 11 open-market sell(s) vs 0 buy(s) — net distribution
INSTITUTIONS (13F) adding as of 2026-03-31: 128 new / 53 closed positions; 259 increased / 117 reduced; institutional ownership +7.56pp; +86 net 13F holders
MGMT LANGUAGE 5/10 measured Shipped SuperRes and AI-app beta; flagship AI app framed as early with potential, not quantified outcomes.
commit “We recently started early customer access to a private beta testing phase of our new AI app”
commit “This year, we introduced SuperRes.”
hedge “Although the app is still in early testing”
VERBATIM AI QUOTES
“These awards reinforce the U.S. government's commitment to integrating commercial AI-enabled geospatial intelligence into its natural security architectures.”
— William Marshall, Q1 FY2027
“we recently started early customer access to a private beta testing phase of our new AI app, a pioneering tool aimed at making Planet's massive global data archive acquirable through natural language. By leveraging Planet's daily data and integrating large language models, it can help nontechnical users to search the data through space and time conduct complex time series analysis, generate answers and produce automatic insights and even analytic reports at speed and scale.”
— William Marshall, Q1 FY2027
“We also launched a new feature called SuperRes, an AI-powered technology to improve the resolution of our plan scope data into a 2-meter class resolution visual product, providing clarity for human-in-the-loop analysis with speed and frequency.”
— William Marshall, Q1 FY2027
“our investments in AI and position Planet at the forefront of the industry, making planetary scale, insights accessible and actionable to more users than ever before”
— William Marshall, Q1 FY2027
“it also has a satellite to sell like communications as well as the NVIDIA chips that we're flying with the other ones. In combination is what enables the more real-time insights going from hours, latency for getting analysis of after you take a picture to minutes.”
— William Marshall, Q1 FY2027
“the merger of Earth imaging data with AI and the large language models really asking the value latent in earth imaging data in new ways and lowering the barriers to entry”
— William Marshall, Q1 FY2027
“AI is ready to help us tie that quicker, I think. So that's the exciting thing, and it's also where Planet is so uniquely positioned because all of this is because of our daily scan.”
— William Marshall, Q1 FY2027
“the AI solutions that we have, like things like GMS, our MDA as well as this beta of a new natural language-based interface is really enabling us to engage with customers that haven't historically really thought about how they might integrate GIS into GIS data sets into their modeling and analysis”
— Ashley Whitfield Johnson, Q1 FY2027
“we are early days in -- certainly in the AI application but also even in exploring how the solutions that we've initially explored primarily in defense and intelligence, are unlocking opportunities for us in the commercial sector. So we remain very optimistic about how the commercial sector can be a major growth vector for us in the years to come.”
— Ashley Whitfield Johnson, Q1 FY2027
“invested strongly in AI, and announced a cutting-edge partnership with Google to demonstrate satellites for compute in space”
— Will Marshall, Q4 FY2026
“Our integration of Bedrock Research is going very well. The team is helping us scale rapidly and deliver AI-based solutions, notably standing up 600 new monitoring sites within three hours compared to a weeks-long process when we first launched the service.”
— Will Marshall, Q4 FY2026
“exploring the use of NVIDIA's accelerated GPU-based computing platform for Planet Labs PBC data processing, enabling faster, more efficient processing for all of our customers; testing NVIDIA's new Thor processor for in-space use, enhancing super-resolution and other AI processing capabilities”
— Will Marshall, Q4 FY2026
“we are collaborating to build the world's first scaled GPU-native AI engine for satellite data and drive huge advances in efficiency and latency”
— Will Marshall, Q4 FY2026
“while LLMs offer users the incredible ability to have conversations with the text of the Internet, they know very little about the physical world. Real-world models need real-world data, and Planet Labs PBC has it. Our deep data archive, averaging over 3,000 collections for every point in the Earth's landmass, represents a treasure trove for indexing the physical world and training next-generation models. As Wikipedia was the foundation dataset for LLMs, we believe that Planet Labs PBC's Daily Scan is foundational to real-world models.”
— Will Marshall, Q4 FY2026
“AI itself is commoditizing software development, making data the key differentiation in AI.”
— Will Marshall, Q4 FY2026
“we think that more generic AI solutions will soon empower nontechnical users to go from a concept to a bespoke application in under an hour”
— Will Marshall, Q4 FY2026
“last year we saw the start of returns on our investments into Satellite Services. This year, we expect to see the start of returns on our investments in AI.”
— Will Marshall, Q4 FY2026
“Planet Labs PBC is the first space-and-AI company.”
— Will Marshall, Q4 FY2026
“we have seen some early results that are very promising, with big speedups like 100x on certain parts of our codebase. Getting answers to our clients faster is really important.”
— Will Marshall, Q4 FY2026
“Through our AI-enabled solutions, we accelerate time to value, become more deeply embedded in our customer operations, and gain more direct visibility to our customers' operational needs.”
— Ashley Johnson, Q4 FY2026
ANALYST QUESTIONS ON AI
Q (Q1 FY2027, Jeff Van Rhee (Craig-Hallum)): On the AI side, interesting exceeding you've got the beta program now up with the natural language query. Can you just talk about the scope of the trial? How many participants, thoughts on when that goes GA? And I'd love to hear is kind of maybe your mind, what are a couple of more compelling use cases you're seeing people playing with right now?
A: Will Marshall: 'it's very early days... we're in a beta testing mode... we have a cohort of beta-testers just so that we can find where the best value use cases we could dive into... what's tantalizing about it is the Planet historically has faced the solution gap that is that our data in principle can answer a lot of questions. In practice, it's difficult... this direction can help a lot of others by enabling people to be able to build bespoke solutions on top of our data leveraging this kind of technology, which can unlock that for a lot more players... AI is ready to help us tie that quicker.'
Q (Q1 FY2027, Edison Yu (Deutsche Bank)): I wanted to ask you about the [Planetary Intelligence paper] you put out very, very recently... in a more kind of operational sense, what do you think is kind of the next, call it, 2 to 3 years, how do we see that sort of manifest either in the business or industry?
A: Will Marshall: 'that's very much the first part of what I was talking to in that Planetary Intelligence SA, which is the merger of Earth imaging data with AI and the large language models really asking the value latent in earth imaging data in new ways and lowering the barriers to entry... right now, earth imaging data and other sort of space data sets and AI enabling us to do more real-world models, more real-world models open up real-world applications on the ground today in farming, in energy, in insurance and so on. Most of the entities that we serve today are big governments, big enterprise commercial players and this book can enable that to be lower.'
Q (Q1 FY2027, Edison Yu (Deutsche Bank)): In your kind of early work around the engineering [for orbital data center], do you have a sense on what kind of compute density is realistic in the next couple of years?
A: Will Marshall: 'I'm not getting those sort of technical specific to this stage... this is really, as Google put it, a moonshot at the present time... there's several interrelated challenges to do with compute radiators and the interconnect between all the different satellites as well as their computers on board any one satellite... efficiency of chip space is a really important part as well as the networking of those together and the firmware to optimize all of it.'
Q (Q1 FY2027, Colin Canfield (Cantor)): Can you maybe talk about kind of any color on initial discussions that you and Google are having with the chip suppliers? And then maybe talk about kind of what do you think are the key engineering signposts that investors should look for as Planet on-ramps for orbital compute?
A: Will Marshall: 'it's all leveraging their TPU architecture, testing those working space... we're also testing things like inter-satellite links because we will be formation flying these sort of satellites together and other technologies, radiation the radiators for the power... within 10 years, it will definitely be cheaper to do it in space than on the ground... Planet is well positioned because of our history of doing hundreds of satellites... We already do a lot of stuff with AI, as you're all aware.'
Q (Q4 FY2026, Colin Canfield (Cantor)): Can you perhaps update us on the timing and the scaling of both the SunCatcher opportunity as well as sounds like a pretty nascent geo-intelligence platform with NVIDIA?
A: Will Marshall: 'SunCatcher is going well. It is early days... this is about putting their TPUs into space. It is an early tech demo... there is a big potential market there long term. As Sundar put it, I think within ten years, he expects most compute spending to go into orbit... to NVIDIA... more focused on the compute on the ground, how we leverage the GPUs in particular to speed up our data processing pipeline... we have seen some early results that are very promising, with big speedups like 100x on certain parts of our codebase.' Ashley Johnson: 'SunCatcher partnership is structured as an R&D partnership, so it is recognized as contra R&D expense... the NVIDIA partnership, that is really just a research collaboration.'
Q (Q4 FY2026, Edison Yu (Deutsche Bank)): What is the latest status on the Anthropic partnership, and have we kind of progressed further from kind of just testing or early testing the models or the training?
A: Will Marshall: 'we are moving from this world of LLMs that can tell you things about the text of the Internet to how models are increasingly trying to move towards real-world models. And real-world models needing real-world data... we have been building these bespoke solutions... Maritime Domain Awareness solution, the Global Monitoring solution, and the Area Monitoring solution for Civil Government... AI has this potential of making that more generic. That is, anyone can turn up, build their own bespoke application of equivalent fidelity in short order, like maybe within an hour... what we are focused on with those research collaborations is how we can build towards that capability... enabling us to build out that capability to expand the TAM.'
Q (Q4 FY2026, Edison Yu (Deutsche Bank)): To get there, what do you see as the biggest bottleneck or thing we should look out for. Is it a question of just needing more compute? Is it a question of just needing more training?
A: Will Marshall: 'it is complex and evolving in that the space is changing so fast... we are seeing capabilities that just a couple months ago we were not able to do because of the advances in especially coding... There is nothing really standing in the way per se. We have the data. That is the critical ingredient, and it is the differentiating ingredient for AI... AI is commoditizing more the software layer that is making the AI piece—the data piece—most useful for AI... there is nothing holding us back there, and it is moving very fast. And that is why I was saying that I think you are going to start to see this come to fruition this year.'
Q (Q4 FY2026, Caleb Henry (Quilty Space)): What makes 2026 the year you first anticipate seeing a return on investment on AI. Was there more of an 'aha' moment that happened, or is this just the natural evolution of the years of investment?
A: Will Marshall: 'we have had revenue from AI a fair bit before. What I mean is in terms of the big way in which AI can unleash those other market potentials, and I think we are going to really start to see those generic solutions... ways in which anyone can turn up, build an application that is relevant to their needs, and then start getting value. That unlocks other markets that we have been talking about for years latent in our data—energy, insurance, finance, and so on... all the pieces are coming together such that that will come to fruition this year, and you will start to really see that take off.'
Q (Q4 FY2026, Greg Pendy (Clear Street)): Is it kind of that the customers through Anthropic will figure out how to use the data and how valuable it is into their daily workflows, or do you think that you will need some boots on the ground to educate the Civil and Commercial markets?
A: Ashley Johnson: 'we were building out the platform to enable smaller customers to really access the data on a self-serve basis. As we are growing those markets and leaning into the AI that Will highlighted, we will be making some targeted investments in those markets where we are seeing the most traction... we can really show, not tell, in these customer meetings, all the things that you can — all the insights you can extract — from the data to answer their specific questions.'