Temporal Raises $550 Million at a $12.55 Billion Valuation as AI Agents Drive Demand

Temporal, the company behind an open-source engine that keeps long-running software from losing its place when something fails, said on September 14 that it raised $550 million in a Series E round at a $12.55 billion valuation. Lightspeed Venture Partners led the round, with Wellington Management, Growth Equity at Goldman Sachs Alternatives and Tiger Global joining as co-leads. SV Angel and accounts advised by T. Rowe Price also participated, while Andreessen Horowitz, Sequoia Capital, Index Ventures, GIC, Sapphire Ventures and Amplify returned as investors, according to the company.

Related reading: Factory Raises $200M at a $5B Valuation, Tripling Its Worth Since April · Hang Ten Systems Raises $53 Million Second Seed, Bringing Sikka's Startup to $85 Million.

The raise more than doubles the $5 billion valuation Temporal reached in February, when it raised $300 million in a Series D led by Andreessen Horowitz. It brings total capital raised to about $1.2 billion, according to GeekWire. The Bellevue, Washington-based company, founded in 2019, now employs 570 people, roughly double its headcount a year ago.

A valuation that doubled in seven months

Temporal gives its core software away. It makes money through Temporal Cloud, a managed service that runs the open-source engine for customers that would rather not operate it themselves. That commercial platform now serves more than 4,300 paying customers, including OpenAI, Snap, Nvidia, Netflix and JPMorgan Chase, the company said. Annualized revenue run rate has passed $250 million and grew more than 200% year over year, while net dollar retention has stayed above 200% since February.

Those are strong figures for a category, workflow orchestration, that has existed for well over a decade. The valuation works out to roughly 50 times annualized revenue, a multiple that assumes much of the growth continues. Lightspeed partner Anoushka Vaswani framed the bet around the underlying shift in how software is built. "Every team building on AI hits the same wall: the demo is easy, production is hard, because the systems around the models can't handle real-world execution," she said in the announcement. "Most of the market solves that by locking teams into a proprietary stack."

The round had been rumored for weeks. SiliconANGLE reported on August 19 that Temporal was in talks for about $500 million. The confirmed round came in slightly larger, with the valuation more than doubling in the seven months since February.

What durable execution actually does

The phrase Temporal uses for its technology is durable execution, and the mechanics are easier to understand than the marketing language suggests. A developer writes ordinary functions in Go, Java, TypeScript, Python or .NET. Temporal records each step of a workflow in an event history. If a server crashes, a network call times out or a downstream service returns an error halfway through, the platform replays that history and resumes from the last completed step instead of starting over. Retries, timers, signals and waits for human approval are handled by the platform, so engineers do not write the recovery scaffolding by hand.

That matters most for work that spans more than a single request. A payment that waits on a fraud check, an order that crosses several systems, or a model that calls tools over minutes or hours all create many points where something can fail. Temporal's pitch is that the workflow survives those failures and finishes, and that operators can inspect or rewind an execution to see exactly where it went. The company has been building toward this problem for years: co-founders Samar Abbas and Maxim Fateev worked on Amazon Simple Workflow Service, Microsoft's Durable Task Framework and Uber's Cadence before starting Temporal in 2019, as the company describes it.

The open-source project is the distribution channel. Developers adopt it for free inside a company, and the managed service later becomes the paid path once the workload reaches production scale, where reliability and support matter more than the license cost.

Why AI agents push demand

Agentic AI changed the scale of the problem. An assistant that answers a question finishes in seconds. An agent that researches, files, books, codes or negotiates may run for hours, days or longer, and it may pause for approval or input along the way. Each additional tool call is another chance to fail, and rerunning a long agent from the beginning wastes both time and model tokens. Temporal's argument is that the state handling payment systems always needed is now a requirement for agents as well.

The figures published with the round point to that shift. Temporal Cloud processed 1.9 trillion billable actions in August, up more than 350% year over year. Open-source installs passed 43 million, up 134% since January. Snap moves 414 million Stories a day on the platform, and OpenAI's use of Temporal grew 60-fold in under a year, per the announcement. OpenAI's vice president of infrastructure, Venkat Venkataramani, said in the same release that durable execution had become "a core requirement for modern AI systems," and that the reliability it provides was one reason OpenAI built a durable orchestration framework on Temporal.

Temporal has been positioning itself for agents throughout the year. In April it shipped an integration that lets OpenAI's Agents SDK run sandboxed coding agents as durable workflows, so a session can survive a crash, switch infrastructure providers or idle for days without losing its workspace, the company explained. In August it previewed an agent harness that wraps existing SDKs with policies for tool approvals and audit trails, according to Temporal. Abbas described the failure mode in blunt terms: "As agents take on more critical work across more systems, every additional step creates another place to fail. In production, that work has to survive those failures and finish reliably."

The competitive field around agent infrastructure is crowded. Model providers, cloud vendors and a growing list of agent frameworks all want to own the layer where agents plan and act. Temporal's bet is that the durable state layer sits beneath those frameworks and outlasts whichever model or SDK a team picks, which is why the company has integrated with OpenAI's, Google's and other agent toolkits rather than competing with them. The risk is that the winning frameworks bundle enough durability that customers stop paying for a separate platform.

The open-source business, and its limits

Temporal's model rests on the gap between a free project and a paid service. The engine is available under the MIT license, but many enterprises pay for Temporal Cloud rather than run the system themselves, because operating a durable execution platform at scale takes real work. That is a well-tested route to revenue in infrastructure software, and net dollar retention above 200% suggests customers expand their usage over time rather than starting small and leaving.

Competition is real, even if it has not slowed Temporal much. AWS Step Functions and Google Cloud Workflows offer managed orchestration inside their own clouds, and some engineering teams assemble workflow layers from queues and databases. Temporal's counterargument is portability: the same workflow can run on the company's cloud, on a customer's own servers or on another provider, and it can span multiple clouds without being rewritten. The risk is the familiar tension in open-source infrastructure. If the big cloud providers make orchestration good enough and cheap enough to bundle, the paid tier has to justify itself on reliability, visibility and enterprise features that are hard to copy.

A venture market tilted toward AI

Temporal's round landed in a venture market that has become extraordinarily concentrated. AI companies accounted for 86% of U.S. venture deal value in the first half of 2026, about $355.9 billion, according to the Q2 2026 PitchBook-NVCA Venture Monitor. U.S. startups raised more than $400 billion in the same period, surpassing every previous full-year total on record. The concentration runs deep: PitchBook found that the five largest deals in the first quarter alone, among them rounds for OpenAI, Anthropic and xAI, captured nearly three-quarters of all venture investment.

That backdrop cuts both ways for a company like Temporal. Being filed under AI infrastructure is part of what makes a $12.55 billion valuation possible in a seven-month span, because growth funds and their backers are directing capital at anything that supports AI workloads. The same enthusiasm has drawn warnings that spending may be outrunning returns. Reuters noted that the round closed despite concerns over high spending and calls to slow development. Temporal is not a model developer, which insulates it from some of that debate, but its growth is tied to customers that are themselves spending heavily on AI.

What to watch

The first question is whether revenue keeps pace with the valuation. Temporal would need to roughly double annualized revenue again, toward $500 million, to bring its multiple closer to more mature infrastructure companies. The company says it will spend the new money on global expansion, platform research and development, and the reliability and security work enterprises keep requesting.

The second is how durable execution holds up as a category. If agents keep moving into production, demand for state management, retries and audit trails should grow with them, and Temporal's head start plus 43 million open-source installs give it a wide base. If agent deployments stall or get absorbed into cloud platforms, the market contracts. The company's own research, a survey of more than 550 engineers it published in August, found that daily AI agent use rose 70.8% in a year, with 80.8% of respondents using agents daily or more, up from 47.3%, according to the report. It is a vendor's survey and should be read that way, but the direction it describes aligns with what Temporal's own usage numbers show.

The third is the exit question that hangs over every late-stage enterprise software company. At $12.55 billion, Temporal sits squarely in initial public offering territory, and its investor list now includes mutual funds and asset managers of the kind that tend to appear as a company matures. For now the company is choosing to grow privately, with the money to do so.

Cover: Temporal wordmark. Source and license recorded by the TechPastWeek picture desk.