<!-- Canonical: https://www.aigentcy.com/research/malta-ai-adoption/methodology/ -->

Research

# How we built these pages

Every figure comes from 4 Eurostat datasets on ICT use in enterprises, pulled through Eurostat's API on 10 November 2026 and shown as Eurostat published it.

-   **Prasad Subrahmanya**Head of AI and Automation
-   **Simon Bonello**Founder

Published 10 November 2026

## Every figure comes from Eurostat's survey of ICT use in enterprises

Eurostat publishes the results of the EU survey on ICT usage and e-commerce in enterprises. NSO Malta runs the survey in Malta and sends the results to Eurostat, which is the source of record for these pages.

| Publisher | Dataset | Code | Edition | Retrieved | Link |
| --- | --- | --- | --- | --- | --- |
| Eurostat | Artificial intelligence by size class of enterprise | isoc_eb_ai | Updated 15 June 2026; years 2021, 2023, 2024, 2025 | 10 November 2026 | Data browser, DOI |
| Eurostat | Integration of internal processes by size class of enterprise | isoc_eb_iip | Updated 27 February 2026; years 2021, 2023, 2025 | 10 November 2026 | Data browser, DOI |
| Eurostat | Cloud computing services by size class of enterprise | isoc_cicce_use | Updated 27 February 2026; years 2021, 2023, 2024, 2025 | 10 November 2026 | Data browser, DOI |
| Eurostat | Artificial intelligence by NACE Rev. 2 activity | isoc_eb_ain2 | Updated 15 June 2026; years 2021, 2023, 2024, 2025 | 10 November 2026 | Data browser, DOI |
| Eurostat | Structural business statistics, number of enterprises | sbs_sc_ovw | Updated 30 September 2026; year 2024 | 10 November 2026 | Data browser |
| NSO Malta | ICT usage and e-commerce in enterprises, Malta |  | Annual survey, reported to Eurostat |  | nso.gov.mt |

Eurostat data are © European Union, reused under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) as set out in [Commission Decision 2011/833/EU](https://eur-lex.europa.eu/eli/dec/2011/833/oj). The values are unchanged. [Eurostat's survey metadata](https://ec.europa.eu/eurostat/cache/metadata/en/isoc_e_esms.htm) describes the survey.

## The 2026 figures come from a separate Commission survey

The European Commission's Flash Eurobarometer 578, "AI and algorithmic management in European SMEs and mid-caps".

-   **Who.** Companies with 1 to 749 employees, all sectors. 260 interviews in Malta and 12,826 across the EU, by telephone.
-   **When.** 20 March 2026 to 13 April 2026, published 14 October 2026.
-   **What.** It asks about "artificial intelligence or automated systems in the workplace", so it counts automated systems that aren't AI. With 260 interviews, a Malta figure near 40% has a margin of about 6 points either way.
-   **How we use it.** We show it in its own section and never join it to the Eurostat series. The figures come straight from the [Malta factsheet](https://europa.eu/eurobarometer/api/deliverable/download/file?deliverableId=109711) and the [data annex](https://europa.eu/eurobarometer/api/deliverable/download/file?deliverableId=108712). An obstacle's share adds "very" and "rather" important difficulty.
-   **Targets.** Malta's AI targets, 20.7% for 2025 and 27.2% for 2030, are from the [Digital Decade 2026 country report: Malta, European Commission, 17 June 2026](https://ec.europa.eu/newsroom/dae/redirection/document/128655). They use Eurostat's indicator for firms with 10 or more staff.

## 38 indicators, with Eurostat's codes and wording

Our short name for each is the one the tables use.

Use at least one AI technology  
`E_AI_TANY`

Enterprises using at least one of the AI technologies: AI\_TTM, AI\_TSR, AI\_TNLG, AI\_TIR, AI\_TML, AI\_TPA, AI\_TAR, E\_AI\_TPVSG. Dataset `isoc_eb_ai`.

Use no AI technology  
`E_AI_TX`

Enterprises don't use any of the AI technologies: AI\_TTM, AI\_TSR, AI\_TNLG, AI\_TIR, AI\_TML, AI\_TPA, AI\_TAR, E\_AI\_TPVSG. Dataset `isoc_eb_ai`.

Use 2 or more AI technologies  
`E_AI_TGE2`

Enterprises using at least two of the AI technologies: AI\_TTM, AI\_TSR, AI\_TNLG, AI\_TIR, AI\_TML, AI\_TPA, AI\_TAR, E\_AI\_TPVSG. Dataset `isoc_eb_ai`.

Use 3 or more AI technologies  
`E_AI_TGE3`

Enterprises using at least three of the AI technologies: AI\_TTM, AI\_TSR, AI\_TNLG, AI\_TIR, AI\_TML, AI\_TPA, AI\_TAR, E\_AI\_TPVSG. Dataset `isoc_eb_ai`.

Text mining (analysing written language)  
`E_AI_TTM`

Enterprises using AI technologies performing analysis of written language (text mining). Dataset `isoc_eb_ai`.

Speech recognition  
`E_AI_TSR`

Enterprises using AI technologies converting spoken language into machine-readable format (speech recognition). Dataset `isoc_eb_ai`.

Generating written or spoken language or code  
`E_AI_TNLG`

Enterprises using AI technologies generating written or spoken language or programming codes (natural language generation, speech synthesis). Dataset `isoc_eb_ai`.

Image recognition or processing  
`E_AI_TIR`

Enterprises using AI technologies identifying objects or persons based on images (image recognition, image processing). Dataset `isoc_eb_ai`.

Machine learning for data analysis  
`E_AI_TML`

Enterprises using machine learning (e.g. deep learning) for data analysis. Dataset `isoc_eb_ai`.

Automating workflows or helping make decisions  
`E_AI_TPA`

Enterprises using AI technologies automating different workflows or assisting in decision making (AI based software robotic process automation). Dataset `isoc_eb_ai`.

Autonomous robots, vehicles or drones  
`E_AI_TAR`

Enterprises using AI technologies enabling physical movement of machines via autonomous decisions based on observation of surroundings (autonomous robots, self-driving vehicles, autonomous drones). Dataset `isoc_eb_ai`.

Generating pictures, video or audio (asked from 2025)  
`E_AI_TPVSG`

Enterprises using AI technologies generating pictures, videos, sound/audio. Dataset `isoc_eb_ai`.

Marketing or sales  
`E_AI_PMS`

Enterprises using AI technologies for marketing or sales. Dataset `isoc_eb_ai`.

Production processes  
`E_AI_PPP`

Enterprises using AI technologies for production processes. Dataset `isoc_eb_ai`.

Business administration or management  
`E_AI_PBAM`

Enterprises using AI technologies for organisation of business administration processes or management. Dataset `isoc_eb_ai`.

Logistics  
`E_AI_PLOG`

Enterprises using AI technologies for logistics. Dataset `isoc_eb_ai`.

ICT security  
`E_AI_PITS`

Enterprises using AI technologies for ICT security. Dataset `isoc_eb_ai`.

Accounting, controlling or finance  
`E_AI_PFIN`

Enterprises using AI technologies for accounting, controlling or finance management. Dataset `isoc_eb_ai`.

Research and development or innovation  
`E_AI_PRDI`

Enterprises using AI technologies for research and development (R&D) or innovation activity. Dataset `isoc_eb_ai`.

Bought ready to use  
`E_AI_ARDY`

Enterprises' AI technologies were commercial software or systems ready to use. Dataset `isoc_eb_ai`.

Built or adapted by an outside provider  
`E_AI_AEXT`

Enterprises' AI technologies were developed or modified by external providers. Dataset `isoc_eb_ai`.

Bought, then adapted by own staff  
`E_AI_AMOWN`

Enterprises' AI technologies were commercial software or systems modified by own employees. Dataset `isoc_eb_ai`.

Open source, adapted by own staff  
`E_AI_AOS`

Enterprises' AI technologies were open-source software or systems modified by own employees. Dataset `isoc_eb_ai`.

Built by own staff  
`E_AI_ADOWN`

Enterprises' AI technologies were developed by own employees. Dataset `isoc_eb_ai`.

Considered using AI  
`E_AI_EC`

Enterprises that ever considered to use one of the AI technologies: AI\_TTM, AI\_TSR, AI\_TNLG, AI\_TIR, AI\_TML, AI\_TPA, AI\_TAR, E\_AI\_TPVSG. Dataset `isoc_eb_ai`.

Costs seem too high  
`E_AI_BCST`

Enterprises do not use AI technologies, because the costs seem too high. Dataset `isoc_eb_ai`.

Lack of relevant expertise  
`E_AI_BLE`

Enterprises do not use AI technologies, because of a lack of relevant expertise. Dataset `isoc_eb_ai`.

Doesn't fit existing equipment, software or systems  
`E_AI_BINC`

Enterprises do not use AI technologies, because of incompatibility with existing equipment, software or systems. Dataset `isoc_eb_ai`.

Data hard to get or of poor quality  
`E_AI_BDDT`

Enterprises do not use AI technologies, because of difficulties with availability or quality of the necessary data. Dataset `isoc_eb_ai`.

Data protection and privacy concerns  
`E_AI_BCDP`

Enterprises do not use AI technologies, because of concerns regarding violation of data protection and privacy. Dataset `isoc_eb_ai`.

Unclear legal consequences  
`E_AI_BLEG`

Enterprises do not use AI technologies, because of a lack of clarity about the legal consequences. Dataset `isoc_eb_ai`.

Ethical concerns  
`E_AI_BEC`

Enterprises do not use AI technologies, because of ethical considerations. Dataset `isoc_eb_ai`.

AI isn't useful for the business  
`E_AI_BNU`

Enterprises do not use AI technologies, because artificial Intelligence technologies are not useful for Enterprise. Dataset `isoc_eb_ai`.

ERP software  
`E_ERP1`

Enterprises who have ERP software package to share information between different functional areas. Dataset `isoc_eb_iip`.

CRM software  
`E_CRM1`

Enterprises using Customer Relationship Management (CRM) software (as of 2023). Dataset `isoc_eb_iip`.

Business intelligence (BI) software  
`E_ITBI`

Enterprises using Business Intelligence (BI) software. Dataset `isoc_eb_iip`.

Any of ERP, CRM or BI  
`E_BSANY`

Enterprises using any business software (any of ERP, CRM, BI). Dataset `isoc_eb_iip`.

Paid cloud computing services  
`E_CC1_PANY`

Enterprises using at least one of the paid CC services: CC\_PEM, CC\_PSOFT, CC\_PFACC, CC\_PERP, CC\_PCRM, CC\_PSEC, CC\_PDB, CC\_PFIL, CC\_PCPU, CC\_PDEV. Dataset `isoc_cicce_use`.

Units: `PC_ENT` is percent of all firms in the size class; `PC_ENT_AI_TANY` is percent of firms that use at least one AI technology; `PC_ENT_AI_TX` is percent of firms that use no AI technology; `PC_ENT_AI_EC` is percent of firms that don't use AI and considered using it.

## How we define staff, AI use, sectors and the gap

Staff

Eurostat counts persons employed: employees plus working owners and family workers. "Firms with 250 or more staff" means 250 or more persons employed.

Firm

An enterprise, the unit Eurostat surveys.

Uses AI

The firm uses at least one of the listed AI technologies (`E_AI_TANY`). In 2025 the list has 8: text mining, speech recognition, generating language or code, generating pictures, video or audio, image recognition, machine learning for data analysis, automating workflows or helping make decisions, and autonomous robots or vehicles. Before 2025 it had 7, without generating pictures, video or audio.

EU

The total for the 27 member states since 2020 (`EU27_2020`), as Eurostat calculates it.

Small member states

Cyprus, Estonia, Ireland, Luxembourg and Slovenia: the 5 member states closest to Malta in population and economy, with the same survey and size classes. Dataset `isoc_eb_ai`, indicator `E_AI_TANY`.

Sector

A NACE Rev. 2 section, for firms with 10 or more staff, from dataset `isoc_eb_ain2`. The chart ranks information and communication, professional, scientific and technical, administrative and support services, real estate, transport and storage, manufacturing, wholesale and retail, construction, hotels and restaurants. Electricity and gas, Water and waste, Repair of computers and communication equipment and ICT sector stay in the CSV only. Malta has a handful of firms in each, and Eurostat flags or withholds their earlier years.

Survey scope

Firms with 10 or more staff in NACE Rev. 2 sections C to N without K, plus group 95.1. That leaves out finance and insurance (K), gambling, arts and entertainment (R), public administration, education and health (O, P, Q), agriculture and mining (A, B), and firms with fewer than 10 staff.

Year

The survey year. The AI questions ran in 2021, 2023, 2024 and 2025. ERP, CRM and BI questions ran in 2021, 2023 and 2025, every other year. Paid cloud figures for Malta exist for 2021, 2023 and 2025.

Gap

Malta minus the EU, in percentage points, for the same indicator, size class and year.

## 7 steps from Eurostat's API to the number on the page

A script runs every step, and automated tests check the result before publishing.

1.  ### Pull the data
    
    A script (scripts/sync-eurostat-adoption.mjs) calls Eurostat's statistics API once per dataset, for Malta and the EU27, and once each for the 5 peer countries and the sector breakdown. It saves every response unchanged as a snapshot.
    
2.  ### Tidy it
    
    It turns the snapshots into one file with a row per figure: dataset, indicator, unit, size class or activity, area, year, value, Eurostat flag and status. It stops if a value has more than 2 decimals or an indicator disappears from Eurostat's list.
    
3.  ### Apply the flag rules
    
    A confidential cell carries no number, even if the API sent one. A low-reliability cell keeps its number with a "(u)" after it. A cell Eurostat hasn't published reads "Not published".
    
4.  ### Count the firms
    
    From Eurostat's business statistics (sbs\_sc\_ovw, 2024), the number of Maltese firms in each size class is summed over the survey's sectors, to show how much one firm moves a Malta percentage.
    
5.  ### Compute the differences
    
    A gap is Malta minus the EU, in percentage points, only when both cells are clean. A change is one year minus another for the same area, on the same rule. A rank among the peer countries orders their clean values for the same size class and year. Pages show them to 1 decimal.
    
6.  ### Draw the charts and tables
    
    Every figure on the pages, in the charts and in the CSVs reads from the tidy file through one module. Pages and charts round to 1 decimal. The CSVs keep the 2 decimals Eurostat publishes.
    
7.  ### Test before publishing
    
    Automated tests compare the headline figures with the raw snapshot, read by separate code that walks the JSON-stat index directly, and check that a confidential or unpublished cell never shows as a number.
    

## Every headline figure matched a second download

-   Headline figures compared with a second read of the raw snapshot in the test suite, and with a second download from Eurostat's SDMX service in CSV format, read by separate code. Every value matched.
-   Parts checked against the indicator that counts them: the share using 2 or more technologies is never above the share using at least 1.
-   Flags counted: of the 1856 cells in our download, 8 carry a flag and 241 have no published value.
-   The sync script can re-fetch and report any cell that has changed since the last pull.

## 1 large firm moves a Malta figure by about 1.1 points

Malta has 92 firms with 250 or more staff and 509 with 50 to 249 staff in the sectors the survey covers (Eurostat business statistics, 2024).

-   If every firm counted equally, 1 large firm would move a Malta percentage by about 1.1 points, and 1 mid-sized firm by about 0.2. Survey weights change the exact amount. The EU totals rest on tens of thousands of large firms, so they move much less.
-   Questions asked only of some firms have smaller bases. The reasons for not using AI go only to firms that don't use AI and considered it. For Malta's large firms in 2025 that's about 18 firms.
-   The 117 firms with 250 or more staff that Eurostat counts for Malta's whole business economy include sectors this survey leaves out, such as finance and gambling. The 92 count adds up only the survey's sectors:

| NACE | Sector | 50 to 249 staff | 250 or more staff |
| --- | --- | --- | --- |
| C | Manufacturing | 73 | 14 |
| D | Electricity and gas | 1 | 1 |
| E | Water and waste | 1 | 2 |
| F | Construction | 43 | 7 |
| G | Wholesale and retail | 111 | 11 |
| H | Transport and storage | 35 | 13 |
| I | Hotels and restaurants | 67 | 10 |
| J | Information and communication | 39 | 6 |
| L | Real estate | 2 | 1 |
| M | Professional, scientific and technical | 50 | 5 |
| N | Administrative and support services | 87 | 22 |
| S951 | Repair of computers and communication equipment | 0 | 0 |
|  | Total in the survey's sectors | 509 | 92 |

Source: Eurostat, sbs\_sc\_ovw, number of enterprises, 2024.

## A confidential cell never shows a number

Eurostat marks some cells with a flag. Each flag has a rule here.

Low reliability (u)

We show the value with "(u)" after it and don't compute differences from it.

Confidential (c)

We show "Confidential" and no number, on the page and in the CSV, even if the API returns one.

Not published

Eurostat has no value for the cell, for example an EU total it hasn't calculated or a question not asked that year. We show "Not published".

Other flags

A break in series (b), estimate (e) or provisional value (p) shows as a letter after the value.

## How the 51.8% figure for large firms reaches this page

The share of Malta's firms with 250 or more staff using AI in 2025, from the API request to the CSV row.

1.  ### Request
    
    [`https://ec.europa.eu/eurostat/api/dissemination/statistics/1.0/data/isoc_eb_ai?geo=MT&size_emp=GE250&indic_is=E_AI_TANY&unit=PC_ENT&time=2025`](https://ec.europa.eu/eurostat/api/dissemination/statistics/1.0/data/isoc_eb_ai?geo=MT&size_emp=GE250&indic_is=E_AI_TANY&unit=PC_ENT&time=2025)
    
2.  ### Cell
    
    Dataset `isoc_eb_ai`, indicator `E_AI_TANY`, unit `PC_ENT`, size class `GE250`, area `MT`, year 2025.
    
3.  ### Eurostat value
    
    `51.75`, no flag. Dataset updated 15 June 2026.
    
4.  ### On the page
    
    51.8%, at the top of the overview and in the trend chart, the size table and the peer chart. The gap to the EU is 51.75 minus 55.03, so −3.3 points.
    
5.  ### In the CSV
    
    The row for 250 or more staff, 2025, in [ai-use-by-firm-size.csv](https://www.aigentcy.com/research/malta-ai-adoption/data/ai-use-by-firm-size.csv).
    

## The survey records use, and nothing on spend or return

-   How much a firm uses AI, what it spends or what it gets back. The survey records use, yes or no.
-   Anything about finance, insurance or gambling firms, or firms with fewer than 10 staff.
-   Which firms answered. The published data are totals by size class, sector or country.
-   Results from other Malta surveys, which use different samples and questions. We don't combine them with these figures.

## Version

Version

1.0 (Eurostat data retrieved 2026-11-10)

Published

10 November 2026

Data retrieved

10 November 2026

Eurostat update, `isoc_eb_ai`

15 June 2026

Eurostat update, `isoc_eb_iip`

27 February 2026

Eurostat update, `isoc_cicce_use`

27 February 2026

Eurostat update, `isoc_eb_ain2`

15 June 2026

Next refresh

When Eurostat publishes the next survey year. Each change is noted here.

## How we help

1.  [01**AI workflow automation**Invoices, requests and reminders handled in the ERP, CRM and help desk you run.](https://www.aigentcy.com/services/process-automation/)
2.  [02**AI integration services**AI that works on live data in the systems your team already uses.](https://www.aigentcy.com/services/ai-integration/)
3.  [03**AI consulting and enablement**A plan tied to this year's targets, and hours saved with the AI licences you already pay for.](https://www.aigentcy.com/services/ai-consulting/)

## Using this research

Cite as

Subrahmanya, P. and Bonello, S. (2026). AI use in Maltese companies, 2025 and 2026. Aigentcy. https://www.aigentcy.com/research/malta-ai-adoption/methodology/ Data: Eurostat, isoc\_eb\_ai, isoc\_eb\_ain2, isoc\_eb\_iip and isoc\_cicce\_use.

If you use our figures, charts or data, credit Aigentcy and link to this page.

Download the data

-   [Share of firms using AI, by firm size, Malta and EU, 2021 to 2025 (CSV)](https://www.aigentcy.com/research/malta-ai-adoption/data/ai-use-by-firm-size.csv)
-   [Share of firms using AI in Malta, Cyprus, Estonia, Ireland, Luxembourg, Slovenia and the EU, by firm size, 2021 to 2025 (CSV)](https://www.aigentcy.com/research/malta-ai-adoption/data/ai-use-small-states.csv)
-   [Share of firms with 10 or more staff using AI, by sector, Malta and EU, 2021 to 2025 (CSV)](https://www.aigentcy.com/research/malta-ai-adoption/data/ai-use-by-sector.csv)
-   [Every figure on these pages, one row per cell, with Eurostat flags (CSV)](https://www.aigentcy.com/research/malta-ai-adoption/data/all-figures.csv)

Questions about the figures? Email [info@aigentcy.com](mailto:info@aigentcy.com).

## See what AI would change in one of your processes

The 2-week AI readiness assessment measures the hours and volumes in one area of your business and prices what AI could take on there, so the board sees the numbers before anyone commits.

[See the 2-week assessment](https://www.aigentcy.com/ai-readiness-assessment/) [Estimate the hours AI saves](https://www.aigentcy.com/tools/ai-savings-calculator/)

**Sources.** Eurostat, survey on ICT usage in enterprises, datasets [isoc\_eb\_ai](https://ec.europa.eu/eurostat/databrowser/view/isoc_eb_ai/default/table?lang=en), [isoc\_eb\_iip](https://ec.europa.eu/eurostat/databrowser/view/isoc_eb_iip/default/table?lang=en), [isoc\_cicce\_use](https://ec.europa.eu/eurostat/databrowser/view/isoc_cicce_use/default/table?lang=en) and [isoc\_eb\_ain2](https://ec.europa.eu/eurostat/databrowser/view/isoc_eb_ain2/default/table?lang=en), retrieved 10 November 2026. European Commission, [Flash Eurobarometer 578](https://europa.eu/eurobarometer/surveys/detail/3696), 2026. © European Union, reused under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). [Methodology](https://www.aigentcy.com/research/malta-ai-adoption/methodology/).
