During a recent episode of the popular technology and venture capital-focused podcast, “20VC,” an intriguing critique emerged regarding the evolving dynamics of the artificial intelligence training sector. Brendan Foody, the co-founder and current Chief Executive Officer of Mercor, offered candid remarks on his competitor, Scale AI. While acknowledging the accomplishments of Scale AI’s former CEO, Alexandr Wang, Foody was careful to distinguish between areas where Wang demonstrated exceptional skill—particularly in extending the company’s reach through distribution channels and in mastering the complexities of sales—and the domains in which, according to Foody, the company faltered. In his assessment, although Wang’s business prowess was undeniable, Scale AI gradually drifted away from its central priorities. Specifically, Foody emphasized that the company lost its commitment to refining the product itself and to maintaining a governing emphasis on scaling quality, two elements he considered foundational to enduring success in the industry. This shift, he argued, created one of the greatest long-term hurdles the business would confront.

In response to these statements, Scale AI issued a pointed rebuttal through a spokesperson, Joe Osborne. Osborne suggested that if Foody repeatedly references Scale in his commentary, it may be evidence of a kind of admiration, even if couched in criticism, as it garners attention for Mercor by association. He went further to defend Scale’s performance, underscoring that the company’s internal data quality measurements were currently at record-setting levels and reaffirming its singular commitment to market leadership and industry dominance. Mercor itself did not provide any immediate follow-up to Business Insider’s inquiries for additional comments.

The discussion between these rival executives must be situated within the broader landscape of data annotation enterprises. Companies such as Scale AI, Mercor, and Surge AI have carved out a critical role in the artificial intelligence supply chain by employing substantial numbers of part-time contractors worldwide—collectively numbering in the hundreds of thousands. These individuals perform labor-intensive and intellectually demanding tasks: filtering responses, ranking outputs, and refining prompt-driven answers to train models used by some of the largest AI corporations across the globe. Their contributions are essential, as they effectively form the backbone that enables machine learning systems to develop accuracy, reliability, and a level of contextual understanding that approximates human judgment.

Foody, who co-founded Mercor in 2023, repeatedly returned to one dominant thesis: that the ultimate determinant of success in the annotation and AI training industry is quality. He argued that to achieve excellence, companies must not only uphold stringent standards for the output itself but also ensure a culture of respect and equitable treatment for the individuals providing the labor. As an illustration of this philosophy, Foody cited Mercor’s compensation practices, explaining that contractors working for his company are compensated at an average hourly rate of approximately $95. He contrasted this figure with his perception of rival firms: according to him, organizations such as Scale AI and Surge AI typically offer closer to $30 an hour. Of course, these rates are not universal; remuneration often varies depending on a contractor’s geographical location, academic background, or the intrinsic complexity and specialization level of the assignment.

The discrepancies in payment scales are corroborated, to some extent, by public information disclosed on company websites. Scale AI, for example, outlines compensation of roughly $30 to $50 per hour for specialists with STEM expertise and a lower range of approximately $15 to $30 for generalist contractors. By comparison, Mercor advertises substantially higher pay: STEM specialists can reportedly earn between $90 and $110 per hour, while generalists are paid around $45. In June, further clarification of Mercor’s approach was provided by its head of product, Osvald Nitski. He revealed that Mercor’s broader organizational strategy extends beyond compensation to a discerning recruitment strategy. The company deliberately seeks annotators of extraordinary caliber—individuals such as International Mathematical Olympiad medalists, Rhodes Scholars, and doctoral candidates—thereby creating a workforce that embodies both intellectual rigor and distinguished academic achievement.

Foody’s commentary carries heightened significance when considered against the background of Scale AI’s recent organizational turbulence. In June, the company secured a monumental $14.3 billion investment from Meta, positioning it as one of the most heavily backed firms in the space. However, this infusion of capital brought with it intensified scrutiny from investors and large-scale technology clients, who expressed concerns about whether confidential client information could remain protected amid such close integration with Meta. The anxiety was only compounded by revelations published by Business Insider: internal investigations revealed that Scale AI had been using openly accessible Google Docs to manage sensitive projects for influential clients, including Google, Meta, and Elon Musk’s xAI. Some of these documents bore the label “confidential” yet remained viewable to anyone who possessed the link, highlighting significant vulnerabilities in the company’s security practices. After the issue surfaced, Scale AI asserted that it took the matter seriously and promptly restricted public access to those documents.

The company’s restructuring challenges did not end there. By July, internal pressures led to a round of layoffs that reduced its workforce substantially. Around 200 full-time employees—approximately 14 percent of the total 1,400 staff—were let go, alongside around 500 contractors. Jason Droege, Scale’s interim CEO, sought to present the decision as a necessary though painful corrective measure. In an internal communication obtained by Business Insider, he acknowledged that, while the layoffs seemed appropriate at the time, they inadvertently gave rise to inefficiencies. The company, he admitted, had created a labyrinth of bureaucracy, unnecessary managerial layers, and ultimately, confusion regarding the organization’s overarching direction. This acknowledgment laid bare the extent to which operational missteps had compounded the company’s difficulties, reinforcing Foody’s claim that an overemphasis on growth and distribution can easily cause a venture to lose its grounding in product development and quality assurance.

Taken together, these accounts illustrate the degree to which leadership philosophy and workforce management shape success in the rapidly expanding field of AI training. While Mercor emphasizes elite recruitment, exceptional pay, and a devotion to quality, Scale AI’s recent trajectory suggests a company grappling with the consequences of scaling too quickly without adequately safeguarding the fundamentals. Foody’s commentary therefore resonates not only as competitive positioning but also as a broader lesson for the industry: in cutting-edge technology, the ultimate determinant of enduring success may lie less in aggressive expansion than in thoughtful stewardship of talent and an unrelenting focus on building a product of uncompromising quality.

Sourse: https://www.businessinsider.com/scale-ai-lost-focus-product-quality-mercor-ceo-brendan-foody-2025-9