Racconta il software che entra nei sistemi fisici e regolati: energia, robotica, spazio, difesa, chirurgia, flotte, quantum hardware, engineering industriale.

Octopus Energy sembra una utility, ma la parte piu' importante e' Kraken: un sistema operativo per utility globali nato dentro il retail energy. PhysicsX porta modelli fisici dentro l'engineering industriale. ICEYE, NewOrbit e Skyfora trasformano spazio e reti telecom in infrastruttura di sensing. NEURA Robotics, Theker e Sereact spostano l'AI dentro robot capaci di agire in ambienti reali. Alta Ares, CameraMatics e Uncovr rendono osservabili workflow critici.

Il pattern e' chiaro: la prossima generazione di startup europee non vende solo applicazioni.

Costruisce sistemi operativi per mercati complessi.

L'Europa compete quando trasforma mercati fisici e regolati in piattaforme software.

Energia retail.

Margini sottili.

Regolazione pesante.

Clienti sensibili al prezzo.

Billing complicato.

Switching doloroso.

Customer service storicamente pessimo.

Stack legacy.

Eppure proprio questa complessita' e' diventata il laboratorio operativo di Kraken.

Octopus entra nel mercato con una promessa consumer semplice: energia piu' verde, esperienza migliore, tariffe piu' intelligenti. Ma sotto la superficie costruisce un sistema operativo interno per gestire cio' che le utility tradizionali faticavano a modernizzare: billing, customer care, smart meter data, pricing dinamico, device domestici, EV charging, heat pumps, batterie, flessibilita' e virtual power plant.

La tesi e' potente perche' rovescia la sequenza tipica del software enterprise. Octopus non ha costruito un tool in astratto per venderlo alle utility. Ha prima gestito una utility. Poi ha trasformato le operations ripetute in piattaforma.

Il workflow che Kraken sostituisce e' il sistema operativo quotidiano di una utility.

Ogni cliente genera contratti, consumi, letture smart meter, bollette, richieste di supporto, cambi tariffari, dati di device, interazioni di rete, vincoli regolatori e casi anomali. In una utility tradizionale, questi elementi spesso vivono in sistemi separati, lenti, costosi e difficili da modificare.

Kraken unifica questo livello operativo. Non e' solo billing. Non e' solo CRM. Non e' solo data platform. E' un layer che collega cliente, consumo, asset, flessibilita' e automazione.

Questo diventa ancora piu' importante quando l'energia passa da flusso unidirezionale a sistema distribuito. Case con batterie, EV, pompe di calore, pannelli solari e smart meter non sono piu' solo punti di consumo. Diventano nodi attivi.

La virtual power plant residenziale di Kraken supera i 2 GW con oltre 500.000 device connessi. Qui il software non gestisce solo bollette. Coordina domanda, device e capacita' energetica distribuita.

Il moat di Kraken nasce da una combinazione difficile da replicare.

Primo: proof operativo dentro Octopus. Il software e' stato stressato da un business reale, con clienti reali, regolazione reale e margini reali.

Secondo: scala. Al FY25 Kraken gestisce 44,2 milioni di account live e oltre 73 milioni contracted. Questa distinzione conta: i live accounts mostrano utilizzo operativo, i contracted accounts mostrano pipeline e domanda futura.

Terzo: referenze enterprise. EDF, E.ON, Origin Energy e Tokyo Gas non comprano software mission-critical per moda. Lo adottano se riduce rischio operativo, accelera migrazioni e regge complessita'.

Quarto: regolazione incorporata. In energia, compliance, privacy, resilienza finanziaria, auditabilita', smart meter data e AI governance non sono dettagli. Sono parte del prodotto.

La distribuzione di Kraken e' delicata perche' vende a incumbent che possono vedere Octopus come concorrente.

Per questo lo spin-off e' strategico. Non e' solo una mossa finanziaria. Serve a costruire neutralita' percepita, separazione dei dati e fiducia commerciale verso utility che non vogliono dipendere direttamente da un competitor retail.

Kraken vende migrazione, riduzione del rischio e modernizzazione operativa. Questo significa cicli di vendita lunghi, contratti enterprise, proof tecnici, supporto al change management e forte credibilita' regolatoria.

Ma una volta dentro, il prodotto diventa profondamente integrato. Non sostituisce una dashboard. Sostituisce il modo in cui una utility opera.

Il posizionamento di Octopus/Kraken e' forte perche' separa due economie diverse.

Il retail energy resta un business a margini sottili. I numeri FY25 lo mostrano bene: 13,7 miliardi di sterline di ricavi, gross margin all'8,4%, underlying EBITDA di 90 milioni e net loss di 255 milioni.

Kraken, invece, e' l'asset software scalabile: 422 milioni di sterline di contracted ARR e una valutazione dichiarata di 8,65 miliardi di dollari.

Questa separazione spiega il caso. Octopus usa un mercato difficile per costruire credibilita' software. Kraken usa quella credibilita' per vendere infrastruttura agli incumbent globali.

La utility diventa laboratorio. Il software diventa prodotto. La compliance diventa moat.

Il vantaggio AI di Kraken non nasce da un chatbot.

Nasce dalla densita' operativa dei dati: consumi, tariffe, smart meter, device, customer service, asset domestici, flessibilita', anomalie, switching, pagamenti e vincoli regolatori.

Questi dati hanno valore perche' sono collegati ad azioni reali. Non descrivono solo intenzioni. Descrivono consumo, prezzo, comportamento, capacita' distribuita e risposta agli incentivi.

In un mercato energetico sempre piu' volatile, il vantaggio non e' solo sapere cosa consuma un cliente. E' orchestrare milioni di micro-decisioni: quando caricare un EV, quando usare una batteria, come gestire una heat pump, come spostare domanda, come mantenere affidabilita' e compliance.

Qui l'AI diventa utile solo se e' inserita in un sistema operativo affidabile.

Il timing e' favorevole per quattro ragioni.

Primo: la transizione energetica rende la rete piu' complessa, distribuita e intermittente.

Secondo: smart meter, EV, batterie e heat pumps trasformano le case in asset energetici controllabili.

Terzo: gli incumbent devono modernizzare stack legacy senza interrompere servizi critici.

Quarto: la regolazione spinge verso trasparenza, resilienza, protezione dati e controllo dei sistemi AI.

Octopus/Kraken si trova all'incrocio di queste curve. Non vende solo software a utility. Vende capacita' operativa per un mercato energetico che non puo' piu' essere gestito con strumenti vecchi.

La lezione per i founder e' molto concreta: se vuoi costruire software per un mercato complesso, a volte devi prima vivere il problema come operatore.

Octopus non ha osservato le utility dall'esterno. Ha costruito una utility, affrontato clienti, bollette, regolatori, margini, device, dati e incidenti operativi. Solo dopo ha trasformato quella conoscenza in Kraken.

Questo e' un playbook riusabile:

entra in un mercato dove il software legacy crea frizione reale.

Usa il servizio come laboratorio.

Trasforma operations ripetitive in prodotto solo dopo proof operativo.

Costruisci neutralita' se devi vendere agli incumbent.

Fai della compliance una feature, non un costo a valle.

La parte piu' importante e' questa: nei mercati regolati, il software migliore non promette semplicemente efficienza. Riduce rischio, aumenta controllo e rende possibile una nuova forma di operation.

This week is not about AI that writes better.

It is about software entering physical and regulated systems: energy, robotics, space, defence, surgery, fleets, quantum hardware and industrial engineering.

Octopus Energy looks like a utility, but the important asset is Kraken: an operating system for global utilities born inside retail energy. PhysicsX brings physical models into industrial engineering. ICEYE, NewOrbit and Skyfora turn space and telecom networks into sensing infrastructure. NEURA Robotics, Theker and Sereact move AI into robots that can act in real environments. Alta Ares, CameraMatics and Uncovr make critical workflows observable.

The pattern is clear: the next generation of European startups is not only selling applications.

It is building operating systems for complex markets.

Europe competes when it turns physical and regulated markets into software platforms.

EP56 is the most useful case of the week: Octopus Energy should not be read only as a green utility, but as a tech company that used retail energy to build Kraken.

Kraken manages billing, customer care, smart meters, dynamic tariffs, home devices, EV charging, flexibility and virtual power plants. By FY25 it managed 44.2 million live accounts, more than 73 million contracted accounts and 422 million pounds of contracted ARR. The announced spin-off, with a reported 8.65 billion dollar valuation, also helps make Kraken more neutral for incumbent utilities such as EDF, E.ON, Origin and Tokyo Gas.

The lesson: in regulated markets, the strongest software often comes from real operations.

PhysicsX raised a 300 million dollar Series C at a valuation of roughly 2.4 billion dollars to accelerate physics AI for industrial engineering.

The point is not simply making simulations faster. The point is changing the workflow through which companies in aerospace, defence, semiconductors, automotive, energy and data centers design complex physical systems.

If a simulation that used to take hours or days becomes almost interactive, engineers can explore thousands of configurations, anticipate physical constraints and reduce the cost of error. Here AI does not replace engineering. It expands the capacity to design.

ICEYE raised 450 million euros to scale sovereign satellite intelligence. NewOrbit closed 16 million euros for commercial satellites in very low Earth orbit. Skyfora raised 6.5 million euros to turn telecom networks into a global atmospheric sensing platform.

These are three versions of the same need: seeing better, more often, with less dependence on external infrastructure.

ICEYE turns SAR observation into an asset for defence, insurance and emergency management. NewOrbit bets on lower orbits to improve resolution and latency. Skyfora uses existing telecom infrastructure to create denser, real-time weather data. The signal: operational sovereignty starts with data about the world.

NEURA Robotics announced up to a 1.4 billion dollar Series C to scale cognitive robots and physical AI. Theker raised 85 million dollars for AI-native industrial robots. Sereact, after its 110 million dollar round, is targeting the US market with Cortex 2 and more than 1 billion picks in production.

Industrial robotics is changing nature. It is no longer enough to program a machine to repeat a stable task. Customers want systems that can see, adapt, manipulate variable objects and work inside real environments.

The message: the value is not only in the robotic arm. It is in the perception, data, model, deployment and operational learning stack.

Alta Ares raised 50 million euros for AI-powered air defence systems. CameraMatics secured up to 49 million euros for AI-powered fleet intelligence. Uncovr closed 7 million dollars to turn surgical video and intraoperative data into structured clinical documentation. UnGuess acquired Testbirds, creating a European group with more than 20 million euros in revenue and 1 million active users.

Defence, commercial fleets, surgery and digital quality are distant markets, but they share one problem: complex, high-risk events with rich data that is hard to use.

AI matters here when it makes observable what used to remain scattered: threats, driving behavior, surgical procedures, bugs, UX, accessibility, security and digital product quality.

Octopus Energy is compelling because it starts from a market few founders would choose to build software in.

Retail energy. Thin margins. Heavy regulation. Price-sensitive customers. Complicated billing. Painful switching. Historically poor customer service. Legacy stacks. And yet that complexity became the operational laboratory for Kraken.

Octopus entered the market with a simple consumer promise: greener energy, better experience, smarter tariffs. But under the surface it built an internal operating system to manage what traditional utilities struggled to modernize: billing, customer care, smart meter data, dynamic pricing, home devices, EV charging, heat pumps, batteries, flexibility and virtual power plants.

The thesis is powerful because it reverses the usual enterprise software sequence. Octopus did not build an abstract tool to sell to utilities. It first operated a utility. Then it turned repeated operations into a platform.

The workflow Kraken replaces is the daily operating system of a utility.

Every customer generates contracts, consumption, smart meter readings, bills, support requests, tariff changes, device data, grid interactions, regulatory constraints and edge cases. In a traditional utility, these elements often live across separate systems that are slow, expensive and hard to modify.

Kraken unifies this operating layer. It is not only billing. It is not only CRM. It is not only a data platform. It is a layer connecting customer, consumption, assets, flexibility and automation.

This becomes even more important as energy moves from a one-way flow to a distributed system. Homes with batteries, EVs, heat pumps, solar panels and smart meters are no longer only consumption points. They become active nodes.

Kraken's residential virtual power plant exceeds 2 GW with more than 500,000 connected devices. Here software does not only manage bills. It coordinates demand, devices and distributed energy capacity.

Kraken's moat comes from a combination that is hard to replicate.

First: operational proof inside Octopus. The software was stress-tested by a real business, with real customers, real regulation and real margins.

Second: scale. By FY25 Kraken managed 44.2 million live accounts and more than 73 million contracted accounts. This distinction matters: live accounts show operational usage, contracted accounts show pipeline and future demand.

Third: enterprise references. EDF, E.ON, Origin Energy and Tokyo Gas do not buy mission-critical software because it is fashionable. They adopt it if it reduces operational risk, accelerates migration and handles complexity.

Fourth: embedded regulation. In energy, compliance, privacy, financial resilience, auditability, smart meter data and AI governance are not details. They are part of the product.

Kraken's distribution is delicate because it sells to incumbents that may see Octopus as a competitor.

That is why the spin-off is strategic. It is not only a financial move. It helps build perceived neutrality, data separation and commercial trust with utilities that do not want to depend directly on a retail competitor.

Kraken sells migration, risk reduction and operational modernization. That means long sales cycles, enterprise contracts, technical proof, change-management support and strong regulatory credibility.

But once inside, the product becomes deeply integrated. It does not replace a dashboard. It replaces how a utility operates.

The Octopus/Kraken positioning is strong because it separates two different economies.

Retail energy remains a thin-margin business. FY25 numbers show it clearly: 13.7 billion pounds in revenue, 8.4% gross margin, 90 million pounds in underlying EBITDA and a 255 million pound net loss.

Kraken, by contrast, is the scalable software asset: 422 million pounds in contracted ARR and a reported 8.65 billion dollar valuation.

That separation explains the case. Octopus uses a difficult market to build software credibility. Kraken uses that credibility to sell infrastructure to global incumbents.

The utility becomes the laboratory. The software becomes the product. Compliance becomes the moat.

Kraken's AI advantage does not come from a chatbot.

It comes from the operational density of the data: consumption, tariffs, smart meters, devices, customer service, home assets, flexibility, anomalies, switching, payments and regulatory constraints.

This data matters because it is tied to real actions. It does not only describe intent. It describes consumption, price, behavior, distributed capacity and response to incentives.

In an increasingly volatile energy market, the advantage is not only knowing what a customer consumes. It is orchestrating millions of micro-decisions: when to charge an EV, when to use a battery, how to manage a heat pump, how to shift demand, how to maintain reliability and compliance.

AI becomes useful here only when embedded in a reliable operating system.

The timing is favorable for four reasons.

First: the energy transition makes the grid more complex, distributed and intermittent.

Second: smart meters, EVs, batteries and heat pumps turn homes into controllable energy assets.

Third: incumbents need to modernize legacy stacks without interrupting critical services.

Fourth: regulation pushes toward transparency, resilience, data protection and control of AI systems.

Octopus/Kraken sits at the intersection of these curves. It is not only selling software to utilities. It is selling operating capability for an energy market that can no longer be managed with old tools.

The founder lesson is very concrete: if you want to build software for a complex market, sometimes you first have to live the problem as an operator.

Octopus did not study utilities from the outside. It built a utility, faced customers, bills, regulators, margins, devices, data and operational incidents. Only then did it turn that knowledge into Kraken.

This is a reusable playbook:

enter a market where legacy software creates real friction.

Use the service as a laboratory.

Turn repeated operations into product only after operational proof.

Build neutrality if you need to sell to incumbents.

Make compliance a feature, not a downstream cost.

The most important part is this: in regulated markets, the best software does not simply promise efficiency. It reduces risk, increases control and enables a new form of operation.