22-year-old founders approach $10bn valuation connecting domain experts to AI giants as training becomes the new bottleneck

THE NEWS: What Happened

Three Harvard dropouts in their early twenties are reportedly on track to create a $10bn company by solving artificial intelligence’s most human problem: finding qualified experts to train increasingly sophisticated models. Mercor, founded in 2022 by Thiel Fellows Brendan Foody, Adarsh Hiremath, and Surya Midha, is in discussions for a Series C funding round that would value the AI training startup at over $10bn, up from its $2bn valuation achieved just seven months ago during a $100m Series B led by Felicis.

The company has become the connective tissue between AI giants and domain specialists, providing scientists, doctors, lawyers, and other experts to companies including OpenAI, Meta, Google, Microsoft, and Amazon for model training and refinement. Mercor claims to be approaching $450m in annualised run-rate revenue—a six-fold increase from the $75m ARR reported in February—while generating $6m in profit during the first half of 2025. The rapid growth reflects the AI industry’s insatiable demand for human expertise as model capabilities expand into specialised domains.

Multiple investors have reportedly approached Mercor with preemptive offers, with at least two attempting to raise special purpose vehicles to participate in the anticipated round. The startup faces increasing competition from Scale AI (which has sued Mercor for alleged trade secret theft), Surge AI (reportedly seeking a $25bn valuation), and the possibility that clients like OpenAI might develop internal recruiting capabilities following their recent launch of an AI-powered hiring platform.

THE INTELLIGENCE: What It Means

Mercor’s meteoric valuation trajectory illuminates a fundamental paradox in artificial intelligence: the most sophisticated automated systems require unprecedented human input to function effectively. The startup’s potential $10bn valuation—representing a 400% increase in seven months—suggests investor recognition that AI training infrastructure represents one of the sector’s most defensible and scalable business models.

The competitive dynamics reveal an emerging oligopoly in AI training services. With only a handful of companies capable of providing the specialised talent required by major AI laboratories, market power is concentrating among players who can recruit and manage domain experts at scale. Mercor’s reported concentration of revenue from clients including OpenAI creates both opportunity and risk—while large contracts drive rapid growth, customer concentration could prove problematic if AI giants develop internal capabilities or negotiate more favourable terms.

The economic structure deserves particular scrutiny. Unlike typical software businesses that scale through code replication, Mercor’s model requires human labour for each additional dollar of revenue. The company’s reported profitability despite rapid growth suggests either exceptional operational efficiency or pricing power that allows substantial margins above contractor costs. However, this labour-intensive approach may limit long-term scalability compared to pure software solutions.

The regulatory and legal environment adds complexity to competitive positioning. Scale AI’s lawsuit against Mercor alleging trade secret theft reflects intensifying competition for both talent and customer relationships in a rapidly expanding market. Such legal challenges could constrain hiring practices and increase operational costs, though the outcome remains uncertain and may ultimately strengthen intellectual property protections across the sector.

The age and background of Mercor’s founding team presents both advantages and vulnerabilities. Their status as Thiel Fellows and Harvard dropouts provides credibility within Silicon Valley networks and suggests strong pattern recognition capabilities. However, managing a $10bn enterprise with complex client relationships and regulatory challenges typically requires seasoned leadership—addressed partly through the recent appointment of former Uber executive Sundeep Jain as president.

The timing appears optimal for aggressive expansion. As AI models advance beyond general capabilities toward domain-specific applications in healthcare, law, finance, and science, the demand for qualified training experts will likely accelerate. Companies that establish dominant positions in expert recruitment and management may capture disproportionate value as AI capabilities expand into regulated and specialised sectors requiring deep subject matter expertise.

THE BRIDGE: What To Do About It

For venture capitalists evaluating AI infrastructure opportunities, Mercor’s funding round highlights several critical investment themes. The most compelling opportunities appear to be companies that provide essential services to AI development workflows, particularly those creating network effects or switching costs that prevent customer defection to internal alternatives or competitor platforms.

Similar opportunities requiring immediate attention:

• Specialised data providers: Companies aggregating proprietary datasets for AI training in regulated industries like healthcare, finance, or legal services where data quality and compliance create natural moats
• AI evaluation and testing platforms: Startups providing systematic assessment of model performance across domains, particularly those developing industry-specific benchmarking standards
• Compliance infrastructure for AI: Companies helping AI developers navigate regulatory requirements across jurisdictions, especially those with deep expertise in emerging AI governance frameworks

Active investors in AI training infrastructure:

• Felicis: Demonstrated conviction in AI labour marketplaces through Mercor investment, likely seeking similar network effect businesses in AI development stack
• Andreessen Horowitz: Significant AI infrastructure focus with portfolio including companies across the AI development lifecycle from training to deployment
• Sequoia Capital: Multiple AI infrastructure investments including Scale AI competitor, suggesting ongoing interest in AI training and data preparation services
• Google Ventures and Microsoft Ventures:
•  Strategic investors with direct insight into AI development bottlenecks and potential acquisition interest in breakthrough solutions

For founders targeting AI infrastructure markets, Mercor’s trajectory provides strategic guidance about capturing value in rapidly evolving ecosystems. The key insight involves identifying human-intensive bottlenecks in AI development workflows and building scalable solutions that create network effects or switching costs. The emphasis on domain expertise suggests opportunities in verticals requiring specialised knowledge rather than general AI capabilities.

Revenue concentration risk deserves careful attention for AI service providers. While large contracts from major AI laboratories enable rapid scaling, dependence on a small number of clients creates vulnerability to changing procurement practices or internal capability development. Successful companies will likely need geographic and customer diversification strategies alongside defensible competitive positioning.

International expansion presents both opportunity and complexity for AI training companies. Different jurisdictions maintain varying regulations regarding data handling, professional licensing, and cross-border service provision. Companies that develop compliance frameworks for operating across multiple regulatory environments may command premium valuations and capture larger addressable markets.

For corporate development teams at established technology companies, the AI training sector presents acquisition opportunities as internal AI initiatives require external expertise and infrastructure. Companies with existing professional networks or domain expertise in specific verticals may find strategic value in acquiring AI training capabilities rather than building competing platforms internally.

The broader implications extend beyond AI training to any professional services sector where artificial intelligence creates new demand for human expertise. Industries including management consulting, legal services, and technical writing may experience similar dynamics as AI capabilities expand into domains requiring human judgment, creativity, or domain-specific knowledge that cannot be easily automated.