Pomegra
London's Arrakis raises $38M from Blossom Capital and Accel at a $140M valuation, deploying autonomous AI agents for workers in energy, aerospace, and logistics.
Key Takeaways:
- Arrakis closed a $30M Series A led by Blossom Capital on top of a $7.5M Accel seed, reaching $38M raised in six months of operation.
- Datadog founder and CEO Olivier Pomel and OpenAI's head of business products Olivier Godement each participated as angels.
- The company claims customers have cut procurement cycle times by 90%, with production deployments measured in weeks rather than months.
Lead
London-based Arrakis emerged from stealth on July 22, 2026, having raised $38 million to deploy AI agents directly inside energy, aerospace, logistics, and manufacturing operations. The $30 million Series A, led by Blossom Capital, follows a $7.5 million seed round that Accel led in March. The post-money valuation sits at $140 million, reached six months after the company's founding in January 2026.
The round attracted two notable angels: Olivier Pomel, founder and CEO of Datadog, and Olivier Godement, OpenAI's head of business products. Junaid Hussain, founder of Cambridge Aerospace, also participated.
What Does Arrakis Actually Build?
Arrakis deploys AI agents into mission-critical industrial operations, the step most enterprise AI projects never reach. The company describes its model as forward-deployed AI engineering: teams embed at customer sites and get agents into production in weeks, not the months or years typical of legacy enterprise software vendors. Early customers span NYSE-listed companies across energy, logistics, and industrial sectors.
The founding thesis runs on a straightforward numbers argument. Most AI investment has targeted knowledge workers, roughly 30% of the global workforce. The 70% operating factories, supply chains, flight operations, and energy grids have been largely bypassed. Arrakis is betting the financial return from that underserved majority dwarfs anything available in the productivity software market.
Why Are Investors Backing Industrial AI Now?
The short answer is deployment economics. Industrial operations generate vast amounts of structured and semi-structured data - maintenance logs, procurement records, sensor feeds - but have historically lacked tooling to act on that data in real time. The arrival of capable large language models, combined with improvements in agentic frameworks, has lowered the barrier to domain-specific agents that work within these environments without requiring years of custom integration.
For Blossom Capital and Accel, the bet is that industrial AI faces less incumbent competition than enterprise software more broadly. Legacy vendors have dominated operational systems for decades but have been slow to offer production-grade agentic tooling. A startup built from scratch around the agent paradigm can, in theory, move faster and price aggressively.
The presence of Pomel and Godement signals something more specific. Pomel built Datadog by embedding deeply with engineering teams and selling on measurable infrastructure outcomes. That playbook - land with a narrow, provable use case and expand as the data flywheel matures - maps directly onto what Arrakis is describing.
The Team's Pedigree
Founder Rafael Quintanilla spent years as a vice president at Accel before building Arrakis, putting him on both sides of the table for this round. Co-founders Haroun Beltaifa and Romain Fouilland both came from Palantir, which has spent a decade doing forward-deployed AI work with defense and industrial customers. Mikhail Galkov previously engineered at Delivery Hero. Together they bring direct experience in both enterprise data infrastructure and the operational complexity of industrial environments.
That background matters because industrial AI deployments fail most often not on model quality but on integration - the messy work of connecting agents to existing systems, ensuring data quality, and getting operational teams to trust automated outputs. Palantir alumni have seen that failure mode at scale and built careers around avoiding it.
Outlook
Arrakis enters a crowded but still-early field. Competitors including Palantir, Sight Machine, and a growing list of vertical AI startups are pitching the same buyers. The $140 million valuation implies investors believe Arrakis can win on speed and deployment model rather than proprietary model technology, a reasonable bet if the forward-deployed approach generates the rapid proof points that feed enterprise sales cycles. The company has not disclosed revenue figures or customer counts, so the 90% procurement cycle reduction claim remains a single data point without broader context. The Series A gives it enough runway to build that evidence base, or not.
