Restate Raises $20M to Take On Temporal in Agent Durability

Restate has raised a $20 million Series A led by Singular, with Redpoint Ventures and Capital One Ventures participating, TechCrunch reported on September 30. The Berlin startup builds durable execution infrastructure — and it did not set out to serve AI agents at all.

Co-founder Stephan Ewen started the company in 2022 to make multistep workflows survive crashes and network interruptions. "It never was built for agents in the beginning, but it just happened to be a perfect match for all these problems that agents surface," he told TechCrunch.

Why agent workloads landed here

The fit is structural. Agent workflows run longer than traditional software and take inherently unpredictable paths, which makes replay and reproducibility the hard part rather than raw throughput. Ewen's framing: "You need to make sure you track exactly what you do to make it reproducible and have a consistent outcome."

That demand has shown up in revenue. Restate has closed multiple six- and seven-figure customer contracts over the last few months, according to Ewen, and has picked up vibe-coding platform Replit as a customer. Beyond AI companies, he said the startup serves Fortune 500 customers including in financial services — all with the same requirement, that a system failing mid-process automatically reassembles itself.

Restate vs Temporal: the size gap is the story

The capital is aimed squarely at Temporal, the incumbent in durable execution. The gap is not subtle. Temporal was founded in 2019 and announced a $550 million Series E at a $12.55 billion valuation earlier this month — an order of magnitude more capital than Restate's entire round, raised by a company three years older.

Restate's technical differentiator is architectural: instead of building its durable execution engine on top of an external database, it developed its own storage, replication and redundancy layers. Ewen says that makes the engine fast and lightweight. The commercial bet follows from it — durability engines are typically reserved for heavyweight workflows, and Restate is trying to push a more efficient, cheaper system into workloads that would not have justified one before.

Treat the speed claim as the founder's, not a benchmark: no independent numbers accompany it.

The team behind it

The credentials are in stream processing. Before Restate, Ewen co-created Apache Flink, the open source stream-processing framework, then became CTO of Data Artisans, the company commercializing it. Alibaba acquired Data Artisans in 2019 and rebranded it Ververica, where Ewen stayed as CTO for three years.

He founded Restate with former Data Artisans and Ververica colleagues Igal Shilman and Till Rohrmann. Ahmed Farghal, previously an engineer at Stripe and Meta, joined as the fourth co-founder at this Series A.

What the money buys

The plan is go-to-market hiring, more engineers, and an expanded San Francisco Bay Area office to sit closer to current and potential customers — a standard Series A shape for a European infrastructure company chasing US enterprise accounts.

Ewen's longer bet is category expansion: "If we fast-forward a couple of years, it's one of those tools, like a database, that I think pretty much any company will have a use case for, no matter their shape and size."

Worth watching

The interesting question is not whether durable execution matters for agents — teams shipping long-running agent workflows already hit crash recovery and duplicate side effects the hard way. It is whether a $20 million challenger can win evaluations against a far better-capitalised competitor, on an architecture argument about not sitting on an external database. If you are picking a durability layer this quarter, that architectural difference is the thing to benchmark yourself, on your own retry and replay patterns.