Two former early Datadog engineers just raised $7 million to bet that companies won’t want to hand their source code straight to the same AI labs that might one day compete with them. According to TechCrunch AI, the startup is called Niteshift, and its seed round was led by Greylock’s Jerry Chen. That’s a small check by today’s AI standards. The names behind it are not.
What happened
Niteshift, founded by Sajid Mehmood (CEO) and Conor Branagan, came out of stealth with a seed round and a roster of heavyweight angels. As TechCrunch AI reports, that list includes Reid Hoffman, Datadog co-founders Olivier Pomel and Alexis Lê-Quôc, Braintrust’s Ankur Goyal, and Misha Laskin of Reflection AI.
The quick version:
- Who: Two early Datadog engineers who helped scale the company to a multi-billion-dollar valuation.
- What: An “AI coding cloud” that routes between models (GPT, Claude, open source) instead of locking customers into one.
- How it charges: Like a cloud provider, with per-minute usage rates. Not by the token.
- The pitch: Reduce your dependence on the frontier labs, don’t replace your coding agent.
The core bet
Mehmood’s argument is built on a pattern he watched up close. At Datadog, a big chunk of the multicloud business came from e-commerce companies that refused to run on AWS, because Amazon was busy putting many of those same retailers out of business. He thinks the AI version of that story is already playing out.
“We are absolutely going to see the same dynamic as Anthropic goes to compete in legal and healthcare and finance and whatever else,” Mehmood told TechCrunch AI.
The fear he’s selling against is simple: model makers like OpenAI and Anthropic keep launching apps that flatten the startups building on top of them. Some people are calling it the “SaaSpocalypse.”
So Niteshift’s promise is separation. Keep the coding model in one box, keep the orchestration that vets, tests, and maintains AI-generated code in another, and use a vendor that has no competing product to push.
That framing is what won over Chen. “As the frontier labs move up the stack, there’s an opportunity to offer customers an alternate path, unbundling their agents from the infrastructure they run on,” he told TechCrunch AI.
Why it matters
What stands out here is the business model, not just the philosophy.
“Everybody else is selling labor replacement intelligence,” Mehmood said. “We’re selling software to agents, as opposed to humans, but we’re still out here selling software.”
That’s a real distinction. Most AI coding companies price like they’re renting you a developer. Niteshift wants to price like AWS, charging for infrastructure your agents consume. If autonomous coding agents become normal inside large engineering orgs, the company betting on usage-based infrastructure could sit in a durable spot.
The catch
Model independence isn’t a new idea, and Niteshift is walking into a crowded, well-funded room. TechCrunch AI lays out the competition:
- Cursor: The breakout coding tool, reportedly a possible SpaceX acquisition target.
- Cognition: Fresh off a $1 billion raise at a $26 billion valuation.
- Amazon Bedrock: The incumbent cloud play.
- OpenRouter: The AI gateway that just raised $113 million at a $1.3 billion valuation.
Against all that, Niteshift’s main edge is the founders’ scar tissue. Mehmood and Branagan lived the exact scaling pain that large engineering teams now face with AI-written code, and they’re betting that buyers want infrastructure built by people who’ve actually done it at production scale.
What to watch
The thesis is sound and the timing is sharp, but a $7 million seed against billion-dollar competitors means execution is everything. The signal to track is whether enterprises genuinely treat “don’t lock me into one model” as a buying requirement, or just a nice-to-have they’ll trade away for the convenience of an all-in-one agent. Full details are at the original TechCrunch AI report.