iMerit raises $20M Series B for its platform that helps enrich and annotate large datasets that power advanced algorithms in fields such as machine learning
iMerit, a Los Gatos, CA-based data annotation and enrichment company, raised $20m in Series B funding.
Context & Ripple Effects
When iMerit raised this $20M Series B in early 2020, data annotation was still a back-office category — the Los Gatos company pitched enrichment tooling for generic machine learning pipelines. The related coverage shows how far the category has traveled since: SuperAnnotate's $36M Series B in late 2024 repositioned labeling as fine-tuning and evaluation infrastructure, not rote tagging.
The clearest signal of where the market went is Mercor, which moved from a profitable $32M Series A jobs marketplace to reporting $614M in H1 2026 gross revenue with roughly 91% coming from foundation model makers like OpenAI and Anthropic — and is now discussing new funding near a $20B valuation. iMerit's raise reads as an early entry point into what became the training-data supply chain.
First-order effects
- iMerit gains $20M to scale its annotation and enrichment platform, competing directly with better-funded peers like SuperAnnotate for ML teams' dataset work.
- Buyers — companies building advanced algorithms — get another capitalized vendor in a category where tooling depth, not headcount alone, is becoming the differentiator.
Second-order effects
- As foundation model makers concentrate their data spending on a few large suppliers (Mercor's revenue base shows the pattern), mid-size annotators like iMerit face pressure to specialize in domain expertise rather than compete on commodity labeling volume.
- Capital follows the concentration: rounds in this category have grown from tens of millions (SuperAnnotate, iMerit) to nine-figure raises across the AI data and infrastructure stack, raising the bar for what a Series B must claim.
Third-order effects
- If the pattern holds, expert-labeled data solidifies as a structural layer of the AI capital stack — vetted human labor marketplaces become core suppliers to foundation labs, with valuations set by lab procurement budgets rather than traditional SaaS metrics.
- That dependence cuts both ways: vendors whose revenue is overwhelmingly concentrated among a handful of model makers inherit those customers' training-cycle volatility as a business-model risk.
The trend: Training-data annotation has evolved from outsourced back-office work into a strategically funded layer of AI infrastructure, with supplier consolidation tracking foundation model makers' procurement.