ProgramDefense

The invisible Arsenal. How much Italy can produce, and who knows it

An open program on a single question: how much Italy’s defense supply chain can produce in ninety days, with which components and from where. Six sourced articles, three tools that run in the browser, two forms to fill in and an open case with data. Everything downloadable without leaving an email address.

The thesis

Three lines, to be verified

  1. 01

    The bottleneck of European rearmament is the factory, and the Italian factory is made of SMEs whose numbers nobody adds up: the capacity that is missing is largely capacity nobody sees.

  2. 02

    The cycle shortens first with data and only then with machines. The capacity map is built one node per plant, with data that stays owned by the company and a buyer who sees how much, when and from where.

  3. 03

    That map is the most sensitive information in the system. It must stay in the country: inspectable code, company-owned nodes, a federated layer. The four sovereignty tests apply to any vendor, Yempik included.

The numbers

Six numbers, each with its source

These are the numbers the series rests on. Every value links to its primary source; totals that remain government claims do not appear.

4 versus 2
million artillery shells a year, Russia versus the European Union, according to Commissioner Kubilius
Source: EU Perspectives, April 2026
3 times
the European capacity declared by the Commission in 2024 compared with industry’s estimate
Source: RFE/RL
35%
maximum share of non-EU components in EDIP-funded projects
Source: Council of the EU, December 2025
5,000
suppliers measured by Leonardo on AirSupply, 87% of them SMEs
Source: SupplyOn
7 days
the redesign cycle of a drone in Ukraine, with makers up from 7 to 500
Source: Exponential View
1 year
devoted to data consolidation by the Ministry’s 2026 AI and Defense Strategy
Source: Agenda Digitale
The series

Six pieces, in reading order

Every piece closes with something to do on Monday. Read them in order or start from the one you need.

  1. 01European rearmament is won in the factoryFour million versus two, the money that is there and the capacity nobody sees. Three moves for an SME in six months.Viewpoint · 12 min
  2. 02What AI can do without a byte leavingFour information classes, the management charter, the one place. With the grid and the NIS2 questionnaire to download.Guide · 11 min
  3. 03The invisible supply chainThe prime sees you, you do not see your suppliers. The EDIP paradox and the federated layer, one node per plant.Viewpoint · 10 min
  4. 04The seven-day cycleWhat Ukraine teaches manufacturers: the edge lies in the cycle. Seventeen ways of writing six problems, and the checklist.Viewpoint · 7 min
  5. 05How much can you produce in 90 days?The question your customer will ask in 2027. The calculator, the six data points to answer within an hour, the survey.Guide · 7 min
  6. 06The map cannot live in DenverIndustrial data sovereignty with the numbers: the facts from filings, the word “enterprise”, the four tests.Viewpoint · 8 min
The tools

To use on Monday, without leaving an email address

Three tools run in the browser and send no data. The two documents are PDFs to print and fill in. Answers to the survey and the scale are anonymous.

Calculator90-day calculatorSeven numbers: how many parts in 90 days and which constraint binds, the machines or the supplier.Survey90-day surveyTen anonymous questions on supply-chain visibility. The first dataset the Ministry can read.Self-assessmentArsenal scaleTwelve questions, four levels: where your factory stands in its relationship with data.PDFClassification gridThe four information classes, the allowed tools for each and the eight lines of the management charter.PDFTwo-sided NIS2 questionnaireSide A: the four questions for the AI software supplier. Side B: the form the supplier answers with.
The open case

Vertesa: five tables, eleven hidden problems

Vertesa is the fictional company of the whole series: precision mechanics, 120 people, three products, twenty-four suppliers. We publish its synthetic data, suppliers, bill of materials, non-conformities, orders and production, with eleven problems we have seen in real companies. The exercises are split by function; the teacher’s guide is available on request.

Open the Vertesa case
Three lists of three

What to do, depending on where you sit

If you run an SME

  1. 01Classify information into four classes and write the management charter, in one afternoon with management.
  2. 02Build the single-source supplier map in a sheet, then move it into the one place where know-how and rules live.
  3. 03Measure the cycle between report and decision, and keep the six data points ready to answer “how much in 90 days” within an hour.

If you lead a district or a cluster

  1. 01Distribute the survey to your companies: whoever distributes it receives their group’s results.
  2. 02Ask for the federated layer: one node per plant, data that stays with the companies, a buyer who sees the status.
  3. 03Use the Vertesa case to train purchasing, quality and production on the same data.

If you work for the Ministry

  1. 01The federated production capacity map, one node per plant, before any central platform.
  2. 02Open architectures and the four sovereignty tests written into tenders, for any vendor.
  3. 03Read the survey data before deciding where the incentives go.
Method and limits

What we count, what we do not

  • Every number in the series links to its primary source in the text. Totals that remain government claims, such as Ukrainian drone production, are flagged as such and not used.
  • Vertesa is a fictional company and its data is synthetic, generated with a fixed seed. No data, document or name of a real client appears in the program; the supply-chain companies we met are not named.
  • Companies named are named for public facts, filed accounts and public records, with no judgment on the product. The piece on industrial data sovereignty applies the four tests to Yempik too.
  • The survey is anonymous: no name, no company, no production data. Results are published in aggregate form from forty answers onwards.
  • The Arsenal Index, a reading of the filed accounts of 34 supply-chain companies, and the Report are in preparation and will be published only with verified data. The program is not legal advice and does not discuss weapon systems.
Changelog

What is here, what is coming

v0.1September 15, 2026
  • Six articles in the series, with sources.
  • 90-day calculator, 90-day survey, Arsenal scale.
  • Classification grid and two-sided NIS2 questionnaire as PDFs.
  • Vertesa open case with five tables and the exercises.
In preparation
  • Arsenal Index: 34 supply-chain companies read from their filed accounts.
  • Report v1 with the first forty survey answers.
  • External review of the method by an independent reader.

Contribute to the first dataset: three minutes, no name.

Ten anonymous questions on the visibility of your supply chain. With a hundred answers it becomes the chapter the Ministry can read. If you would rather talk, write to us: we sell nothing on the first call.

Go to the survey or write to us
Keep reading

See also

EU data sovereigntyThe general thesis on the company brain that should not live in a US vendor.AI governance for SMEsThe written rules the management charter starts from.Yempik Public BrainThe graph of all our content, with the path “From the defense factory to its supply chain”.