Arrakis Exits Stealth With $30 Million to Put AI Agents Into Industrial Operations

For two years, the money and mythology of artificial intelligence have chased the knowledge worker — the analyst drowning in spreadsheets, the coder shipping features, the marketer generating copy. Arrakis, a London- and Paris-based startup that stepped out of stealth on July 31, is making the opposite bet: that the largest returns from AI will come not from the desk but from the factory floor, the power grid and the shipping yard.

On that day the company announced a $30 million Series A led by Blossom Capital, with participation from existing investor Accel, which had led its $7.5 million seed. The round brings Arrakis's total funding to roughly $38 million barely six months after the company was founded, and, according to Fortune, values the business at about $140 million post-money. The investor roster reads like a who's who of the software economy: Datadog founder and CEO Olivier Pomel, OpenAI head of business products Olivier Godement, and Cambridge Aerospace founder Junaid Hussein all put in personal money.

The pitch: agents for the 70 percent

Arrakis describes itself as an "AI deployment company" that installs AI agents into the mission-critical operations of industrial enterprises — companies in energy, logistics, aerospace, telecommunications, manufacturing and construction. Rather than ripping out decades-old software and forcing multi-year migrations, the firm says it drops agents into existing systems, letting customers train, deploy and scale custom agents "in weeks, not quarters." Its platform is model-agnostic, so a client is not locked into a single foundation-model provider, and it pledges never to use customer data or IP for third-party model training.

The framing is deliberately populist. "Most AI investment to date has targeted the 30% of workers behind a desk," said co-founder and CEO Rafael Quintanilla in the announcement. "The real ROI lies in the 70% running industrial operations. We're building from Europe for the world, with the ambition to build the Operating System our industrial champions need to win."

Quintanilla is himself a former Accel investor who left venture capital to start the company in January 2026. He assembled a founding team — Haroun Beltaifa, Romain Fouilland and Mikhail Galkov — drawn from Palantir, Delivery Hero, Revolut, Datadog and chip-equipment maker ASML. That pedigree matters to the pitch: Palantir built its business on "forward-deployed engineers" who embed with customers, and Arrakis is borrowing the same playbook, pairing on-site engineers with applied research tuned to industrial workflows.

Early traction and a results-based model

The most striking claim is not the money but the speed. Within six months of launch, Arrakis says it has signed enterprise customers including NYSE-listed firms across energy, logistics and industrial sectors, and has already landed its first US customer, reporting results such as a 90 percent reduction in procurement cycle times.

Perhaps the most consequential detail is commercial: Arrakis ties "a significant proportion" of its fees directly to customer outcomes. In an enterprise-software market long criticized for selling shelfware — licenses bought and never fully used — an outcomes-based model is a bold way to signal confidence, and a hard promise to keep at scale.

Its backers are betting the approach travels. "While many companies are focused on AI applications at the edge of the enterprise, Arrakis is tackling some of the most complex operational challenges facing large industrial organisations," said Ophelia Brown, managing partner at Blossom Capital. Accel partner Sonali De Rycker, who backed the seed, added: "Rafael's vision of how AI will transform the physical economy and his deep understanding of the challenges industrial companies face means he and the team are building a transformative company."

Why heavy industry is the next AI battleground

Arrakis is riding a genuine macro wave. Across the US and Europe, governments are pouring capital into reshoring and supply-chain resilience, yet the enterprises meant to deliver that growth often run on fragmented, decades-old systems. That gap — operational data trapped in software that cannot use it — is exactly what agentic AI promises to close.

But mission-critical is the operative and the perilous phrase. An AI agent that hallucinates a slide deck is an annoyance; one that misroutes a fuel shipment, misconfigures a telecom network or errs in an aerospace supply chain is a safety and liability event. Selling autonomy into environments where failure is measured in blackouts or grounded aircraft demands a level of reliability, auditability and human oversight that general-purpose chatbots have not had to meet. Arrakis's emphasis on customer control, data sovereignty and outcome-tied pricing is, in part, a response to that skepticism.

The company also faces a crowded field. Palantir has spent years courting industrial and defense clients, incumbents like Siemens and Schneider Electric are embedding AI into their own stacks, and vertical AI startups are targeting individual sectors. Arrakis's differentiation — model-agnostic, fast-to-deploy, outcomes-priced — is credible but not yet proven at scale.

What to watch

The next twelve months will tell whether Arrakis is a category or a moment. Watch three things: whether it converts its NYSE-listed pilots into large, multi-year contracts as it triples headcount and expands across Europe, the US and the Middle East; whether the outcomes-based pricing survives contact with the messy realities of industrial rollouts; and whether the 90-percent efficiency claims hold up beyond a handful of showcase customers. If they do, the thesis that AI's biggest payoff lies on the factory floor rather than the office desk will have found its first flagship. If they don't, Arrakis will be an early, well-funded cautionary tale about how hard it is to put autonomous software in charge of the physical world.