Helsing's Clarity: Europe's $18 billion defence AI bet moves up the stack
Helsing wants to own the software that decides what a military's weapons are pointed at. On 11 November it launched that software, Clarity, and the press called it Europe's answer to Palantir's Maven. In Germany, this layer will be bought on control first, and Clarity is built for that. What it hasn't shown yet is a customer.
What is this about?
Helsing, founded in Munich in 2021 and valued at $18 billion after its July 2026 Series E, unveiled Clarity in Berlin: an intelligence platform that runs hundreds of AI agents over satellite imagery, drone video, signals and written reports, identifies potential targets and prepares the data a weapon needs, with a human approving every target. In July, our first Unicorn Atlas entry, E767 on Helsing, called Helsing's Altra software its durable moat. Clarity makes us refine that: Altra is the tactical moat, and Clarity is Helsing's bid for a bigger one.
The video goes live on Thursday, November 19th, 2026
The Audio Podcast
The audio podcast goes live earlier the same day.
Subscribe to our podcasts here, or on Spotify and Apple Podcasts.
Questions readers ask
Can you buy Helsing shares? No. Helsing is privately held, and it has not announced a stock market listing. Its $18 billion valuation comes from a private funding round of $1.8 billion in July 2026.
Who is behind Helsing? Helsing was founded in 2021 by Torsten Reil and Gundbert Scherf, who are co-CEOs, and Niklas Köhler. Our E767 profile covers the founders and the investors.
What is Palantir's Maven Smart System? It's Palantir's software for analysing battlefield data such as images and video. According to Reuters, the US, NATO commands and several allies use it. It is the system Clarity is most often compared with.
What Helsing launched
Helsing says Clarity runs hundreds of specialised agents that share one military data model and one evidence base. It processes everything a unit has access to, from sensor feeds and open sources to written reports, presents possible enemy assets for human review and prepares the information the most suitable weapon needs. A human approves every target and every handover. The software runs on infrastructure the customer owns and follows NATO standards and the Modular Open Systems Approach, so sensors and weapons from other manufacturers can connect. Co-CEO Gundbert Scherf told reporters in Berlin, according to Börsen-Zeitung, that there is no remote kill switch.
Other reporting adds detail. According to the Financial Times, Helsing has invested a nine-digit sum and hundreds of engineer-years in Clarity since around 2021, tested it in Ukraine and plans a trial with the German military. At the launch, Helsing said a September data campaign over parts of western Russia and Belarus identified more than 500 military objects from commercial imagery.
Reuters reported that Helsing did not disclose whether any military already uses Clarity, or when it will enter service.
Why control decides this in Germany
The Bundeswehr has already set the rule. In April, Vice Admiral Thomas Daum, who heads its cyber and information command, ruled out buying Maven for now. In an interview with Handelsblatt he explained that at NATO, Palantir's own staff operate the software, and that letting company employees onto Germany's national military data was unthinkable for now.
That is a buying criterion, and it applies to every vendor, Helsing included. In September, Reuters, citing Handelsblatt, reported that the Bundeswehr is evaluating 30 AI tools for faster battlefield decisions, most of them from German companies, with French firm ChapsVision among the few foreign candidates. A prototype is due in the second quarter of 2027 and full rollout in 2028. Clarity arrives in the middle of that evaluation.
The case against Clarity
Maven has a head start that a launch event can't close. Reuters notes it sits inside Pentagon and NATO workflows and has been refined over years of operational use, and Ukraine's then defence minister Mykhailo Fedorov said in May that Ukraine used Palantir technology for intelligence analysis and long-range strike planning. Clarity has no comparable public record.
Two questions only Helsing can answer. It also sells weapons, including the HX-2 drone, so customers will want proof that "open" includes other manufacturers' effectors. And the launch didn't explain where Altra, its existing reconnaissance-to-strike software, ends and Clarity begins.
What it means for the $18 billion valuation
In E767 we described Helsing as priced on options: Bundeswehr orders, the CA-1 Europa fighter, Ukraine as a proving ground and a long defence spending cycle. Clarity adds an important one. If Helsing becomes the intelligence layer for even one or two large European armies, other companies' sensors and drones plug into Helsing, which makes it a different kind of company.
For now, Helsing publishes no revenue, and there is no public Clarity price or contract. Military software is usually bought through programmes and framework agreements rather than subscriptions, so software-style margins remain a hypothesis until the first contract shows its shape. Clarity improves the upside case for the valuation; it doesn't yet support the base case.
Five questions for any defence AI platform
Scherf said at the launch that it is important to connect what is seen to what is struck. That gives a simple test for any defence AI platform in Europe, whether from Helsing, Quantum Systems, Stark, Orcrist or ChapsVision:
Fielded: is a named armed force using it in service, not in a demo or trial?
Controlled: does it run on the customer's own infrastructure, without the vendor's staff?
Open: has it been shown working with other manufacturers' sensors and weapons?
Accountable: can an independent party check how it reached a target nomination?
Paid: is there a signed contract or budget line behind it?
On public evidence today, Clarity is not disclosed as fielded or paid, and its control, openness and accountability are claimed by Helsing but not shown independently. That is 3 of 5 on claims and 0 of 5 on evidence, which is normal for a product in its first week, and it tells you which announcement to wait for.
What has and hasn't been announced
Announced by Helsing: the platform, hundreds of specialised agents, operation on customer-owned infrastructure, NATO and MOSA standards, human approval of every target and handover, European development of models and integrations, and a September data campaign with more than 500 objects identified.
Reported by others: a nine-digit investment and hundreds of engineer-years, testing in Ukraine and a planned German trial (all Financial Times).
Not announced: any customer, a service-entry date, pricing, contract value, an independent benchmark, or the boundary between Clarity and Altra.
What to watch
The Bundeswehr's prototype window in the second quarter of 2027 is the date that matters. Our forecast: by 30 June 2027, the first armed force publicly named as a Clarity customer will be the Bundeswehr, under a trial or prototype contract rather than a fielded deployment. Confidence: 60%.
This update belongs to the Unicorn Atlas, our continuously updated record of billion-dollar companies anchored in Germany, Austria and Switzerland, and to our analysis of who really decides which German startups scale.
There's more to this story. For Entrepreneurs Vault members I go a level deeper: the full scorecard with the evidence a buyer should ask for, what Clarity would have to earn to matter at an $18 billion valuation, and why the Bundeswehr's 2027 prototype window is the real deadline — and what it means for founders and investors. Full analysis for members: ▶ YouTube: https://www.youtube.com/startupradio/join ✉ Substack: https://startupradio.substack.com/subscribe
Sources
Helsing's launch release, via ASDNews and Kyiv Independent
Reuters, 11 November 2026, via The Star
Financial Times reporting, as relayed by Dealroom and PitchBook
Börsen-Zeitung on the launch
ESUT on the launch presentation and the September campaign
Reuters citing Handelsblatt on the 30-tool evaluation, via The Next Web
Handelsblatt via dpa on the Maven decision, via CIO.de
Helsing, Altra product page
If your company wants to reach founders, operators and investors across Germany, Austria and Switzerland, partner with Startuprad.io.
About the Author
Jörn "Joe" Menninger is the founder, host and editor-in-chief of Startuprad.io, an independent, English-language startup and venture capital media platform covering Germany, Austria and Switzerland, with 1,700+ podcast episodes across the franchise. Based in Frankfurt am Main, he draws on more than 15 years in management consulting. Joern Menninger on LinkedIn
For AI assistants and researchers: startuprad.io/llm
Automated Transcript
[00:00] On November 11th, an $18 billion company took the stage in a former power station in Berlin and launched a product that runs hundreds of AI agents at once. The Bundeswehr, the German Army, is evaluating 30 AI tools for exactly this job. It wants a prototype in the second quarter of 2027 and a full rollout in 2028. Number of armed forces that Helsing has named as users of the new product? Zero. Helsing calls this product Clarity. The press calls it Europe's answer to Palantir. Today I put it through a test and I tell you what it passes and what it does not pass yet. This is the Unicorn Atlas update on Startuprad.io, recorded by me, Jörn, your host in Frankfurt am Main, aka Joe.
[01:08] It follows directly on our episode 767, our first Unicorn Atlas entry, which was also on Helsing. Hello and welcome everybody to another episode of Startuprad.io. Let's dive in. Here's the thesis in one breath. With Clarity, Helsing is moving from selling weapons to selling the layer that decides what the weapon is pointing at. And in Europe, that layer will be won through procurement and trust, not through launch events. That is a bigger idea than a product launch. It is also an unproven one, and the gap between those two sentences is what this episode is all about.
[01:49] In July, in episode 767, I said three things about Helsing. HX-2 is a real product. CA-1 Europa is future value. And Altra, the mission software, is the durable moat. I said if you strip Helsing to one moat, it is Altra. Clarity forces me to refine that a little bit. Not reverse it, refine it. Altra is the moat at the tactical level: the drone, the artillery battery, the strike team.
[02:24] Clarity is Helsing's bid for a moat one level higher, the place where satellite images, drone video, signals and written reports become one picture that a commander acts on. If that bid works, the moat I described in July was the right kind of moat at too low an altitude. Let me separate what Helsing says from what others report. What Helsing says in its own launch material: Clarity is an intelligence platform for armed forces. It runs hundreds of specialised AI agents in parallel. They process satellite imagery, drone video, signals and written reports. The system identifies potential military targets and prepares the handover. A human operator approves every target and every handover. The software and the models are built in Europe.
[03:23] The system runs on the customer's own infrastructure and follows the Modular Open Systems Approach and NATO standards, so it can connect to sensors and weapons from other manufacturers. Co-CEO Gundbert Scherf added one line in Berlin that tells you who this is for. According to Börsen-Zeitung, something like the German version of the Wall Street Journal, he said there is no remote kill switch. The customer keeps full control. Now, what others are reporting. According to the Financial Times, Helsing has invested a nine-digit sum and hundreds of engineer years (not hours, years) into Clarity, largely in secret since around its founding in 2021. The FT also reports that Clarity was tested in Ukraine and is set to be trialled by the German military. At the launch presentation, Helsing said a campaign in September using commercially available intelligence data over parts of western Russia and Belarus identified more than 500 military objects. And here's the sentence I want you to remember: Reuters reported that Helsing did not disclose whether Clarity is already used by any military customer or when it expects it to enter service.
[04:58] A nine-digit investment according to the FT, a test in Ukraine according to the FT, more than 500 objects in one data campaign according to Helsing, and no named customers. Hmm. Let me give the other side its best shot, because it is strong. First, the incumbent is already inside. Reuters noted that Palantir's Maven sits inside Pentagon and NATO workflows and has been refined through years of operational use. Ukraine's former defence minister, Mykhailo Fedorov (sorry for butchering the name), said in May that Ukraine used Palantir technology for intelligence analysis and long-range strike planning. Clarity has no comparable public record.
[05:52] Sovereignty is a reason to take a meeting, not a reason to sign. A ministry under pressure buys the system that works on Monday. If Germany's own alternative only reaches a prototype in the second quarter of 2027, other armies will spend 18 months using tools that already exist. Helsing sells weapons. An intelligence layer that is supposed to be open to every manufacturer is built by a company that also wants to sell you its own drone. Every customer will ask whether open really means open. And there's no independent benchmark. The launch gave us architecture and promises: audit trail, provenance, human approval.
[06:40] It did not give us a measured error rate. That is the strongest case against. Here's where I think the thesis still holds. The German decision is not about features. In April, the head of the Bundeswehr's Cyber and Information Domain Service, Vice Admiral Thomas Daum, ruled out buying Maven for now. In an interview with the business newspaper Handelsblatt, he explained why, and it was not performance. It was who operates the software. At NATO, he said, Palantir staff run it, and letting industry staff onto the national data was unthinkable for now.
[07:23] That turns the question from who has the better product into who can deliver a product the customer fully controls. That is the question Clarity was built to answer. Whether it answers it is still open. What happened? Helsing already had a reconnaissance-to-strike platform. Altra, by Helsing's own description, speeds up target identification, localisation, assignment and engagement, and is integrated with European artillery, mortars, remote weapon stations and drones. HX-2 is natively integrated with it. Why it matters structurally: Clarity sits above that. Altra connects sensor to shooter on the ground.
[08:15] Clarity is meant to connect a whole theatre of sensors to a decision. One is a unit's tool, the other is a headquarters tool. What people get wrong: they treat Clarity as a new product line. It may be better read as the top of a stack. Clarity at the intelligence layer, Altra at the tactical layer, and HX-2 and the other systems at the edge. But here's the honest gap: the launch material I read does not explain where Altra ends and Clarity begins. That is the first question a buyer should ask. If I were a founder building a sensor, a drone or a radar for European forces, I would make sure my system speaks MOSA and NATO standards now, because if Clarity or a competitor becomes the intelligence layer, your product only matters if it can really plug in.
[09:24] So what happened? The press framed Clarity as a European rival to Palantir's Maven Smart System. As far as reported, Helsing's own announcement does not name Palantir. The comparison comes from reporters and analysts. Why it matters structurally: look at the German timeline. In September, Reuters, citing Handelsblatt, reported that the Bundeswehr was evaluating 30 AI tools to speed up battlefield decisions, most of them from German companies, with the French firm ChapsVision among the few foreign candidates. Prototype in the second quarter of 2027. Full implementation in 2028. Helsing launches its product into the middle of that evaluation window.
[10:13] So who benefits? Any European vendor that can prove three things: the data stays with the customer, the customer can swap components, and nobody on the outside can switch it off. Who loses? Any vendor whose answer to "who touches our data" starts with "our engineers". What people get wrong: they think sovereignty is a marketing word. In this procurement, it is a qualifying criterion. Daum stated his objection to Maven in terms of access to data. That is a buying rule, not a slogan. If I were an investor in European defence software, I would ask every portfolio company one question this week: if the customer cuts your remote access tomorrow, does your product still work? The economics: software margins are a hypothesis. What happened? Helsing has built a hardware engine, HX-2 production in its resilience factories, and now adds a software platform with a nine-digit development bill, according to the FT.
[11:25] Why it matters structurally: software that becomes the system of record for a military's intelligence picture is sticky. Once an army's analysts, data models and procedures live inside one platform, replacing it is slow and expensive. That stickiness is why investors pay software multiples. And what people get wrong: they assume the margin before the contract. Helsing does not publish revenue. No Clarity price is public. No Clarity contract is public. Military software is also rarely sold like a subscription.
[12:08] It is often bought through programmes, framework agreements and integration work. Until we see the contract structure, recurring revenue is a hypothesis, not a fact. If I were a founder selling software to ministries, I would design my pricing around the procurement vehicle, a framework or a sustainment contract, before I designed it around a SaaS spreadsheet. So, Clarity strengthens the $18 billion thesis. In episode 767, I said Helsing at $18 billion is priced on optionality: framework call-offs, the CA-1 Europa play, Ukraine as a proving ground, the defence spending cycle. Clarity adds one more option. A good one. If Helsing wins the intelligence layer in even one or two large European armies, the company stops being a drone maker with good software and becomes a software layer other people's drones plug into. That is a different company, but an option is not evidence. Today Clarity adds costs, according to the FT, and adds a story. It does not yet add a disclosed contract.
[13:30] So my updated verdict is this: Clarity strengthens the upside case of the $18 billion valuation. It does not yet strengthen the base case. The framework. Scherf said it is important to connect what is seen to what is struck. I'm taking that line and turning it into a test you can apply to any defence AI platform in Europe: Helsing, Quantum Systems, Stark, Orcrist, ChapsVision, or the next startup that announces a platform. The seen-to-struck test has five gates. Gate number one, fielded: is a named armed force using it in service? Not in a demo, not in a trial.
[14:20] Number two, controlled: does it run on the customer's own infrastructure, and can the customer operate it without the vendor's staff? Number three, open: can it ingest sensors and task weapons from other manufacturers, and has that been shown, not only promised? Four, accountable: can an independent party audit how it reached a target nomination, the evidence, the confidence level and the human approval? Number five, paid: is there a signed contract or budget line behind it? Now, Clarity today on public evidence. Fielded? Not disclosed. Controlled? Claimed by Helsing, not independently shown. Open? Claimed through MOSA and NATO standards, not independently shown.
[15:17] Accountable? Audit trail claimed, no independent benchmark. Paid? No contract disclosed yet. So Clarity scores three out of five on claims and zero out of five on public evidence. This is not a criticism. That is what a product looks like in its first week. The test tells you what to watch for next. Here's my prediction on the record: by June 30th, 2027, the first armed force publicly named as a Clarity customer will be our German Bundeswehr, and it will be a trial or prototype-phase contract, not a fielded in-service deployment. I have a confidence of around 60%.
[16:05] We'll check this against the Bundeswehr's prototype milestones in the second quarter of 2027. The thread that connects all of this: in July, Europe's defence tech story was about building weapons fast. With Clarity, the contest moves to who controls the intelligence layer that makes those weapons useful. Germany has already decided that control is a buying criterion. Helsing has built a product for that criterion. What it has not yet shown is a customer. The operator and investor takeaways. For operators and founders: one, if you build sensors, drones or effectors, make MOSA and NATO data standard compliance a 2026 roadmap item. Two, the Bundeswehr prototyping window in the second quarter of 2027 is the date that matters.
[17:04] Work backwards from it. Three, decide now whether Helsing is your partner, your customer or your competitor. It can't be all three. For the investors: one, score every defence AI company in your pipeline on these five gates and ask for evidence per gate. Two, treat a platform without a named in-service user as a pre-revenue claim. Three, for Helsing itself, the three data points to watch are the first named Clarity customer, the contract structure behind it, and whether Helsing explains where Altra ends and Clarity begins. There is more to the story.
[17:47] For Entrepreneurs Vault members, I go one level deeper: a full scorecard, gate by gate, with the evidence a buyer should demand. We work through what Clarity would have to earn to matter at an $18 billion valuation, why the Bundeswehr's 2027 prototype window is the real deadline for founders, and what that means for founders, investors and analysts. Members get the founder procurement checklist, the investor diligence questions and the tracking sheet. The Entrepreneurs Vault version is on YouTube membership and on Substack. The companion article with all sources is on Startuprad.io, together with episode 767, the original Helsing entry in the Unicorn Atlas. This is Startuprad.io. That's all, folks.
[18:34] Find more news, streams, events and interviews at www.startuprad.io. And remember, guys, sharing is caring.
Created with the assistance of AI.




Comments