Scale AI Raises $1.4B Series G at $25B Valuation
Scale AI has closed a $1.4 billion Series G funding round at a $25 billion valuation, led by Accel, reflecting sustained enterprise demand for data labeling and AI model evaluation infrastructure.
Original sourceScale AI announced a $1.4 billion Series G funding round on June 10, 2026, bringing its valuation to $25 billion. The round was led by Accel with participation from existing investors. The company has positioned itself as critical infrastructure for AI development, offering data labeling, RLHF pipelines, and model evaluation services to enterprise clients including major defense contractors and frontier AI labs.
The funding comes as enterprises accelerate AI deployment and discover that model quality is often bottlenecked not by compute but by data quality and evaluation rigor. Scale's Evaluation Platform and its Rapid application have become go-to tools for companies running red-teaming, benchmarking, and safety testing on large language models. That demand appears to be what's driving the round, rather than speculative growth projections.
Scale has navigated some turbulence in recent years, including workforce adjustments and shifting dynamics with major customers like OpenAI, which has built more internal data infrastructure. The $25 billion valuation nonetheless reflects investor conviction that the data and evaluation layer of the AI stack remains durable and difficult to replicate at quality. Whether Scale can maintain margin and defensibility as model providers continue to vertically integrate remains the central question for the business going forward.
Panel Takes
The Founder
Business & Market
“The buyer here is clear — AI labs and enterprise ML teams pulling from model development budgets — and that's the right answer. But the real stress test is what happens when OpenAI, Anthropic, and Google finish building their own internal eval and labeling pipelines, which they're all actively doing. Scale's moat is supposed to be the quality and scale of its human labeler network plus proprietary tooling, but at $25B they need that moat to be deeper than 'we've been doing this longer.' The unit economics only hold if they can keep winning the highest-margin eval work even as commodity labeling gets automated away.”
The Skeptic
Reality Check
“The valuation math here requires believing Scale is irreplaceable infrastructure, not a high-quality vendor that major customers are actively working to replace — OpenAI has been building internal data ops for years and is a direct competitor to Scale's core business, not just a customer. The scenario that kills this isn't a startup competitor; it's vertical integration by the five companies that represent the bulk of Scale's revenue. I'd want to see the customer concentration numbers before treating $25B as anything other than a bet on Scale winning the race to diversify before their anchor clients cut the cord.”
The Futurist
Big Picture
“The thesis Scale is implicitly betting on: that as models commoditize, the evaluation and data quality layer becomes the primary lever for differentiation, and that no single AI lab will trust a competitor to run its evals — creating durable demand for a neutral third-party infrastructure provider. That's a falsifiable and interesting bet. The second-order effect if Scale wins is significant: a company with evaluation data across every major frontier model becomes the de facto benchmark authority for the industry, which is a power position that looks nothing like a labeling vendor. The dependency is that enterprise AI deployment keeps accelerating and that 'good enough' evals from internal teams don't close the gap — both plausible but not guaranteed.”
The PM
Product Strategy
“Scale has done something most AI infrastructure companies fail at: they've built distinct product lines — Donovan for defense, Evaluation Platform for labs, Rapid for enterprise — rather than one platform that tries to serve everyone and serves no one well. The job-to-be-done is crisp for each segment, which is why the revenue is real and not just ARR theater. The risk in the product strategy is that 'data labeling' and 'model eval' are increasingly being treated as the same procurement decision by enterprise buyers, and Scale needs to make sure its expansion motion connects those two surfaces into a coherent workflow rather than two separate vendor relationships.”