Niteshift, an AI coding startup founded by two early Datadog engineers, has raised a $7 million seed round led by Greylock partner Jerry Chen, with a pitch that bets enterprises will refuse to hand their source code directly to the same frontier labs trying to eat their software markets. The round, announced June 10, 2026, also drew checks from Reid Hoffman, Datadog co-founders Olivier Pomel and Alexis Lê-Quôc, Braintrust's Ankur Goyal, and Reflection AI's Misha Laskin. Founders Sajid Mehmood and Conor Branagan are pitching a coding cloud that routes between models rather than locking customers into one vendor.
The $7M check is modest against the comparable rounds reshaping this category. Cognition recently raised $1 billion at a $26 billion valuation. AI gateway platform OpenRouter pulled in $113 million at a $1.3 billion valuation. Niteshift is entering as a seed-stage challenger to companies already valued more than 100 times larger.
Mehmood, who is CEO, frames the wedge through his Datadog years. The monitoring company built a sizable multicloud business by selling to e-commerce customers who refused to run on Amazon Web Services while Amazon was simultaneously gutting their retail businesses. Mehmood argues the same dynamic is rolling through software now, as Anthropic, OpenAI, and others push into legal, healthcare, and finance verticals where their customers operate.
Key facts
- 01Niteshift raised a $7M seed round led by Greylock partner Jerry Chen on June 10, 2026.
- 02Angel investors include Reid Hoffman, Datadog co-founders Olivier Pomel and Alexis Lê-Quôc, Braintrust's Ankur Goyal, and Reflection AI's Misha Laskin.
- 03Founders Sajid Mehmood and Conor Branagan are early Datadog engineers who helped scale the company to a multi-billion valuation.
- 04Niteshift charges per-minute infrastructure rates rather than selling tokens, routing between Claude Code, Codex, and open source models.
- 05Competitors include Cursor, Cognition (recently valued at $26B on a $1B raise), Amazon Bedrock, and OpenRouter ($113M raise at $1.3B).
Niteshift's product is not a replacement for Claude Code or Codex, the two coding agents that currently dominate developer workflows. It sits underneath them. The platform routes between Claude, GPT, and open source models on a per-project basis, presenting itself as model-agnostic infrastructure rather than another agent.
The pricing model is the second distinguishing bet. Niteshift charges per-minute usage rates like a cloud provider rather than reselling tokens. That's a deliberate positioning move — the company is selling infrastructure to the agents, not selling the agents themselves.
Chen, who led the round, said the thesis is about unbundling. He pointed to the same pattern playing out across the AI stack as labs expand from model providers into full application businesses.
The challenge is timing and competitive density. Model independence is not a new idea, and the companies pursuing it have substantial head starts. Cursor — which may soon be acquired by SpaceX, according to reporting around the deal — has spent years building its developer base. Cognition's $26 billion valuation gives it room to underprice. Amazon Bedrock ships model routing as part of AWS. OpenRouter's $1.3 billion valuation reflects how much capital is already chasing the gateway layer.
Mehmood's counter is operational. He and Branagan didn't just study the engineering organization problem — they scaled through it at Datadog, building out the infrastructure that handled production telemetry for thousands of enterprise customers. The pitch to buyers is that AI-generated code will need to be run, tested, and verified autonomously in real production environments, and that requires infrastructure built by people who have done it at scale.
This story builds on a pattern AI Chat Daily covered earlier this month, when cheaper inference started eating frontier workloads at firms like Harvey. The unbundling thesis — that the model is increasingly a commodity input and the moat moves to orchestration, routing, and verification — keeps showing up in seed pitches and enterprise procurement decisions. Niteshift is wagering it shows up faster in coding than anywhere else.
The skeptic's case is that frontier labs are not standing still. Anthropic and OpenAI are bundling more orchestration, observability, and deployment tooling directly into their agent products, narrowing the gap that companies like Niteshift want to exploit. A model-agnostic layer only matters if customers actually value optionality more than the integration depth a single-vendor stack can offer. With $7M in the bank against competitors carrying nine and ten figures, Niteshift has to prove that wedge holds before its runway runs out.
For the AI coding market, the more interesting signal is who wrote the angel checks. Pomel and Lê-Quôc built the playbook for selling neutral infrastructure into engineering organizations that wanted to hedge against their cloud provider. If the same pattern repeats in AI coding — and the multicloud history says it might — the winners won't be the agents themselves. They'll be the layer that lets enterprises swap agents without rewriting their developer tooling. That's the bet Greylock just funded, and it's the bet that decides whether the next Datadog gets built on top of someone else's model.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
