AI Told Former Google Execs to Build Long-Range Shells. The Startup Raised $85 Million
Former Veritone CEO Chad Steelberg built an AI model to find what could influence the war in Ukraine. It advised against yet another drone and recommended long-range shells for 155mm howitzers. That is how Sceptre came about, and Tiberius Aerospace raised an $85 million Series A and is valued above $1 billion. The lesson for founders: AI simulation can help pick a niche, but the shell's claimed performance has not been independently verified.
AI-processed from Wired; edited by Hamidun News
AI in startups usually helps improve an existing product. The story of Tiberius Aerospace shows a different scenario: a neural network model helped decide what to build in the first place. The company was founded by Chad Steelberg, former CEO of the enterprise AI company Veritone, and its strategy is run by Andy Baynes, a former senior Apple executive who came from Google. This week Tiberius announced an $85 million Series A round (the term for a first large venture round), and according to Wired, the company is valued at more than $1 billion.
How AI Chose What to Build
After Russia's full-scale invasion of Ukraine, Steelberg wanted to help but did not know where to start. He did what he does best: he asked his team to build an AI model. The private research lab he founded in 2021 worked with people in France, the United Kingdom, and the United States and experimented with drones, electric vertical takeoff aircraft, and quantum computing. After the invasion, its AI specialists and aerospace engineers were redirected to code and spreadsheets meant to objectively break down the factors of the war and identify what could shift the situation.
The model's first conclusion: improving defense technology is pointless, because that would only entrench the positional deadlock. Steelberg then devised metrics for evaluating the lethality of systems, including cost-effectiveness, and the model ran different weapon types under different conditions. The advice was unexpected: do not make another drone, make artillery munitions with greater range.
That is how Sceptre came about, a shell fired from a 155mm howitzer, the most common gun among those transferred to Ukraine. The company claims it flies farther than existing munitions, can reach supersonic speeds, and reaches altitudes beyond the zone of electronic warfare jamming. Sam Cranny-Evans, an expert at the British think tank RUSI, says range matters: gun crews could operate beyond the reach of Russian drones. In April, Tiberius fired Sceptre from a NATO-standard howitzer for the first time, and it is now refining the system together with the US Army.
GRAIL: The Simulator as a Product
The company turned the original AI system into a product called GRAIL (Generative Real-Time Artificial Intelligence for Lethality). Steelberg calls it the most interesting direction. The platform automates supply chain management and compliance, and it also simulates weapons against different targets under different conditions and weather, for example on an Arctic peninsula. Beyond open data, the team is trying to obtain data from real combat operations, including from Ukraine. The idea is for users to test and refine designs for specific tasks faster, and for licensed weapon projects to be continuously updated under decentralized production.
An "Apple-Style"
Business Model
Baynes, who spent four years under Steve Jobs, calls the approach the "Apple model." Production is outsourced to many third-party contractors. This speeds up development and output, and expensive manufacturing capacity does not sit on the company's balance sheet in peacetime. Baynes wants to tap spare precision manufacturing capacity in the United Kingdom, France, Germany, the Nordic countries, and the United States.
Money is already coming in. The Series A round was led by the European investor Lakestar and Rochefort Asset Management, a national security fund licensed by the US Department of Defense; the Ukrainian fund UA1 also participated. The company announced a $500 million contract with British partners to produce, license, and distribute the munition. It is also developing a missile and expects to generate revenue from it next year.
What It Means
For founders and product managers, there are two lessons. First: AI simulation can be used to choose a niche, not just to optimize an existing product. Second: licensing projects plus outsourcing production yields a "light" balance sheet, familiar to software companies.
There are also limitations. The model's conclusions depend on the metrics and data, and Steelberg himself set the metrics. Sceptre's specifications are claimed by the company, and the material contains no independent verification or results from combat use. The valuation above $1 billion and the $500 million contract are also reported by the company itself. Finally, using AI in designing lethal weapons raises ethical and regulatory questions, and the market will have to answer them.
- Nov 21, 2026A swimming team captain handed ChatGPT three scheduled tasks and cut her routine work by a multiple
- Nov 21, 2026MIRA Neural Network Renders Rocket League at 20 FPS, but Ping Hits 460 ms
- Nov 21, 2026Google built a closed copy of the internet with fake banks to train Chrome's AI agents
- Nov 21, 2026MLPerf Training Adds Nvidia Rubin and TPU Ironwood in First Appearance: 153 Systems
Key Questions Answered
How AI Chose What to Build?
After Russia's full-scale invasion of Ukraine, Steelberg wanted to help but did not know where to start. He did what he does best: he asked his team to build an AI model. The private research lab he founded in 2021 worked with people in France, the United Kingdom, and the United States and experimented with drones, electric vertical takeoff aircraft, and quantum computing.
What are the key details on GRAIL: The Simulator as a Product?
The company turned the original AI system into a product called GRAIL (Generative Real-Time Artificial Intelligence for Lethality). Steelberg calls it the most interesting direction. The platform automates supply chain management and compliance, and it also simulates weapons against different targets under different conditions and weather, for example on an Arctic peninsula. Beyond open data, the team is trying to obtain data from real combat operations, including from Ukraine.
What are the key details on An "Apple-Style" Business Model?
Baynes, who spent four years under Steve Jobs, calls the approach the "Apple model." Production is outsourced to many third-party contractors. This speeds up development and output, and expensive manufacturing capacity does not sit on the company's balance sheet in peacetime. Baynes wants to tap spare precision manufacturing capacity in the United Kingdom, France, Germany, the Nordic countries, and the United States. Money is already coming in.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.