Datadog veterans launch AI coding startup Niteshift on a bet against Big AI lock-in

AI coding cause startup Niteshift has raised a $7 cardinal effect circular led by Greylock’s Jerry Chen. That’s a humble sum by AI standards, but the startup, founded by 2 erstwhile aboriginal Datadog engineers, has attracted immoderate big-name angels similar Reid Hoffman, Datadog’s Olivier Pomel and Alexis Lê-Quôc, Ankur Goyal of Braintrust, and Misha Laskin of Reflection AI. 

Founded by Sajid Mehmood and Conor Branagan, who helped turn Datadog from its aboriginal days to a multi-billion valuation, the institution has entered the crowded AI coding abstraction with a compelling idea: Why would immoderate institution spot its astir delicate assets — codification that runs its products — straight to exemplary makers similar OpenAI and Anthropic, fixed that those companies are perpetually “killing” startups and businesses by launching competing apps?

Mehmood, who is CEO, likens it to Datadog’s aboriginal growth, erstwhile the monitoring institution won e-commerce customers who refused to physique connected Amazon Web Services. It was a tenable concern, fixed that Amazon was simultaneously putting galore of those aforesaid retail stores retired of concern successful what became known arsenic the “retail apocalypse.”

The AI equivalent, arsenic Mehmood sees it, is already underway. Anthropic, OpenAI, and others are moving accelerated into vertical bundle markets — what immoderate are calling the SaaSocalypse.

“At Datadog we saw this clearly,” Mehmood said. “A large portion of our multicloud concern came from e-commerce businesses who did not privation to tally connected Amazon, right? … We are perfectly going to spot the aforesaid dynamic arsenic Anthropic goes to vie successful ineligible and healthcare and concern and immoderate else.” 

The stake is that companies volition progressively question infrastructure that separates the coding exemplary from each the different orchestration needed to guarantee AI-generated codification is decently vetted and maintained (and that they’ll privation a vendor without a competing agenda).  

To beryllium clear, Niteshift isn’t replacing Claude Code oregon Codex, the 2 astir fashionable coding agents. It argues that it reduces dependence connected them.

Niteshift’s AI coding unreality volition way betwixt those models — on with open-source options and others — based connected the needs of each project.

“Being capable to power betwixt GPT and unreality models is important,” Mehmood said, “Everybody’s disquieted astir getting stepped connected by these giants.”

That thought is what got Greylock’s Chen to bite. 

“As the frontier labs determination up the stack, there’s an accidental to connection customers an alternate path: unbundling their agents from the infrastructure they tally on,” Chen told TechCrunch. “Niteshift is gathering the level that enables this for coding agents, letting customers put profoundly successful their developer tooling without locking themselves into a azygous exemplary oregon cause vendor.”

More than that, Niteshift isn’t selling tokens. It sells infrastructure, charging similar a unreality provider, with per-minute usage rates. 

“Everybody other is selling labour replacement intelligence,” Mehmood said. “We’re selling bundle to agents, arsenic opposed to humans — but we’re inactive retired here selling software.”

Even so, Niteshift is entering a crowded marketplace of AI coding tools. Model independency isn’t a caller idea, and Niteshift’s competitors person a monolithic caput start. That includes Cursor, though it could soon beryllium gobbled up by SpaceX; Cognition, which conscionable raised $1 cardinal astatine a $26 cardinal valuation; Amazon Bedrock; and AI gateway level OpenRouter, which just raised $113 cardinal astatine $1.3 cardinal valuation. The database goes on.  

Mehmood’s reply to each of that is the founding team’s depth. Mehmood and Branagan didn’t conscionable survey these problems — they lived them, scaling Datadog done the nonstop increasing pains that ample engineering organizations present look with AI-generated code. Teams, helium said, request to run, trial and verify bundle autonomously successful their existent accumulation environments, and they request infrastructure built by radical who’ve done it astatine scale.