Recommended Articles
Fundraising

Trajectory Raises $45 Million Series A at $300 Million Valuation

By Angeled · November 3, 2026 · Fundraising
Trajectory Raises $45 Million Series A at $300 Million Valuation
Trajectory Raises $45 Million Series A at $300 Million Valuation

Trajectory, a startup building a platform to retrain open-source AI models using specialist feedback, raised $45 million in a Series A round at a $300 million valuation, according to TechCrunch reporting. The round was led by Sequoia Capital, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures. The company previously raised a $15 million seed round in May 2026.

Trajectory operates in one of the most important and least visible layers of the AI stack: post-training. Foundation models are initially trained on vast amounts of general data, but raw capability is not enough for specialized work. A model that assists doctors, lawyers, engineers, or financial analysts must be refined on expert knowledge and corrected by people who actually understand the domain. That refinement — often called post-training or continual learning — is what Trajectory sells.

Why Post-Training Matters

The AI industry's first phase was dominated by scale: bigger models, more data, more compute. That phase produced remarkably capable general models, but it also revealed a limitation. General models hallucinate, misunderstand domain conventions, and degrade as the world changes — new regulations, new products, new scientific findings. A model trained in 2025 does not automatically know what happened in 2026.

Continual learning addresses this by keeping models current. Instead of retraining a model from scratch — enormously expensive — Trajectory's approach retrains open-source models using feedback from specialists, improving performance in specific domains and keeping the model aligned with evolving knowledge. This is particularly valuable for enterprises that cannot or will not send proprietary data to a closed model provider, and that want models they can run and control themselves.

A Competitive and Well-Funded Field

Trajectory is not alone. Competitors cited in coverage include Thinking Machines Lab, Applied Compute, and Prime Intellect, all working on aspects of model improvement, training infrastructure, or AI research tooling. The presence of Nvidia as an investor is notable: the chipmaker's venture arm backs companies that drive demand for its hardware, and post-training at scale is compute-intensive work.

The $300 million valuation, just months after a $15 million seed round, reflects how quickly investor enthusiasm can compound in AI infrastructure. It also raises the bar: Trajectory must now demonstrate that its platform delivers measurable improvements for customers, at a cost and speed that beat in-house efforts.

Why It Matters for Founders and Investors

For founders building AI products, Trajectory's rise is a signal about where value is migrating. As foundation models become commodities, differentiation shifts to the data and feedback loops that make a model good at a specific job. Companies that own specialist feedback — expert corrections, domain workflows, proprietary evaluation data — may hold more durable advantages than companies that merely wrap a general model.

For investors, post-training is attractive because it is a picks-and-shovels business: it serves many model builders rather than betting on a single model winner. The risk is competition from the model labs themselves, which increasingly offer fine-tuning and customization as built-in features. Trajectory's bet is that an independent, open-source-focused platform can out-specialize the giants.

Angeled.com connects founders with investors.

Read Next