European manufacturing is at a crossroads: labor shortages, rising costs, and global competition are pushing factories to seek new levers of efficiency. Almetra, a Berlin-based startup previously known as Deltia, offers an answer based on artificial intelligence that runs directly on the shop floor, without cloud connectivity and with a keen eye on privacy. Today the company announces a €16.3 million Series A round led by blisce/ — a French fund whose portfolio includes tech giants — joined by NAP, Merantix Capital, Robin Capital, Underline, Critical Ventures, and a group of business angels.

The funding announcement

The financing, explained co-founder and CEO Maximilian Fischer, comes at a time when many manufacturers know they are losing production capacity but lack the tools to understand where inefficiencies hide. Almetra fills this gap with a platform that unifies video, machine data, IT systems, and operator know-how into a single source of operational intelligence.

How the system works

The technological core is a mix of AI cameras installed above production lines and proprietary models trained for industrial environments. Video is processed locally — either on the camera itself or on an edge node — and converted into structured data: cycle times, output rates, equipment utilization. No complex IT integration is needed: the hardware is already talking. To protect workers, footage is anonymized in real time; only short, randomized snippets are retained for root-cause analysis, while most data stays on-site. This privacy-by-design approach is made possible because the models run entirely locally, gradually adapting to each customer's specific processes.

Edge computing and data sovereignty: why it matters for industry

Almetra’s decision to bypass the cloud has implications that go beyond operational efficiency. For many manufacturing companies, the intellectual property embedded in production data is a strategic asset: handing it over to external servers means exposing it to security and compliance risks. GDPR regulations and data residency policies turn local processing into a competitive advantage, not a limitation. Moreover, edge computing reduces decision-making latency — critical when reacting in real time to a machine stoppage — and cuts bandwidth costs by eliminating the need to transfer continuous video streams to remote data centers. The Almetra case shows that industrial AI is following a path similar to what AI-RADAR observes for on-premise deployment of Large Language Models: control, privacy, and cost predictability become higher priorities than cloud convenience, even if they require more direct hardware management and specific in-house skills.

Outlook: robotics and closed-loop automation

With the new funding, Almetra aims to extend its platform from a mere visibility tool to a full-fledged automation layer for the shop floor. Plans include US expansion and the development of robotics applications in selected environments. The ultimate goal is a closed loop: not only detecting inefficiencies but also acting automatically, with robots driven by the same models that now monitor the lines. If realized, this will push the edge intelligence paradigm even further, where every decision is made at the machine level and data never leaves the factory.