Investors Just Put $750M Into Making AI Agents Reliable
Temporal raised $550M at a $12.55B valuation and Factory raised $200M at $5B. The money is moving from models to the plumbing that keeps agents working.
Two of the largest AI funding rounds of mid-September went to companies that do not build AI models at all. Temporal, which keeps long-running software from failing halfway through, raised $550 million at a $12.55 billion valuation. Factory, which builds AI agents that write code for large companies, raised $200 million at a $5 billion valuation, more than three times what it was valued at in April.
Temporal: the problem nobody sees until it breaks
An AI agent does not answer once and stop. It works through a sequence: look something up, call a service, wait, decide, act again. Any step can fail, because a server times out or a network drops. Without protection, the whole task either stops or, worse, repeats steps it already completed, such as charging a card twice.
Temporal's software provides what it calls durable execution: it records each completed step, so if something fails the work resumes where it left off rather than starting over. It is open source, and its customers include OpenAI, NVIDIA, Netflix and JPMorgan Chase. The company says its annual revenue run rate has passed $250 million, growing more than 200% year on year. The round was led by Lightspeed, Wellington Management, Goldman Sachs Alternatives and Tiger Global.
Factory: agents that write enterprise code
Factory builds autonomous coding agents aimed at large organisations. Its backers include Blackstone, Khosla Ventures and Sequoia Capital, and its total funding now exceeds $400 million. The tripling of its valuation in five months reflects how quickly companies are moving from AI that suggests code to AI that writes and ships it.
Why this is the real story
For two years, the money followed the model makers. These rounds show investors now paying for the layer underneath: the systems that make agents dependable enough to trust with real work. It is the same lesson this month's agent incidents taught from the other direction. Capability is no longer the bottleneck. Reliability and control are.
The money is also concentrated. Crunchbase reports that roughly 88% of AI startup funding this year has gone to companies headquartered in the United States.
Questions You Should Be Asking
- If an AI workflow we run fails halfway through, does it resume, restart, or repeat something it should not?
- Are we paying for a more capable model when our real problem is reliability?
- For AI-written code in our systems, who reviews it before it reaches customers?
- With funding this concentrated, which of our AI suppliers could be acquired or shut down, and what is our exit plan?
What To Watch Next
Whether the major AI labs begin building this reliability layer themselves. If they do, standalone companies like Temporal face their largest customers becoming competitors.
Sources
Ready to implement AI in your business?
Our team builds the AI systems you just read about. Start with a free 30-minute discovery meeting.