Scale AI’s defense contracts and federal data labeling revenue

Defense contract expansion

Scale AI secured a $500 million contract through the Pentagon’s Chief Digital and Artificial Intelligence Office to help warfighters interpret battlefield data. The Pentagon’s Chief Digital and Artificial Intelligence Office signed a $500 million agreement with Scale AI to help warfighters interpret battlefield data, marking a fivefold increase from the $100 million contract the company received in 2025 for integrating generative AI into classified military networks. The firm also works on the Defense Innovation Unit’s Thunderforge program and the Golden Dome homeland defense initiative. These defense contracts drive growth as the company negotiates a tender offer that would value the business at $25 billion. This target exceeds the $14.3 billion valuation from Meta’s 49 percent stake in June 2025. Scale AI also provides data support for the U.S. Air Force, PayPal, and Toyota. The company grew to 300 employees from 100 a year ago after raising $325 million in its latest investment round. CEO Alexandr Wang has led the company to raise more than $600 million in venture capital since its 2016 founding. Scale AI also sells Nucleus, a software package that allows customers to search for mislabeled examples to prevent degradation in algorithm performance. The U.S. Army executes more than 100,000 contracting actions to support 180 missions across 50 countries annually.

The data labeling market

The revenue gap between Scale AI and its competitors shows a changing market. Scale AI expects $100 million in revenue over 12 months, though its sales doubled last year. You should watch how these specialized firms capture specific niches.

Appen reported $230.8 million in group revenue for fiscal year 2025, which grew 4.5 percent when excluding the impact of a Google contract termination. Appen China grew 74.8 percent to 102.9 million dollars, while Appen Global fell 21.1 percent to 127.9 million dollars. Its underlying EBITDA hit $12.2 million with a 5.3 percent margin. Generative AI work reached 44.1 percent of fourth-quarter revenue. Management guides FY26 revenue to between $270 million and $300 million. In contrast, Surge AI reached $1.4 billion in revenue in 2025 using 50,000 vetted experts. Mercor reported $2 billion in annual recurring revenue in 2026 and raised $350 million in a Series C round at a $10 billion valuation in October 2025 using more than 30,000 contractors. Mercor matches domain specialists, including PhDs and lawyers, to model-training and evaluation work. TELUS Digital completed its privatization on October 31, 2025, after buying its remaining shares for $4.50 each in a $539 million deal. Defined.ai reported 65 percent year-over-year revenue growth in 2025. EXL completed its $310 million acquisition of iMerit on August 3, 2026. Sama delivers managed annotation across image, video, 3D point cloud, and sensor data. Labelbox provides an annotation platform for image, video, and text that uses model-assisted labeling to speed throughput.

Operational risks and the labor gap

Scale AI builds its $25 billion valuation on the work of data labelers who stay unaware of which clients use their training data. This labor remains invisible as value flows to the cap table. The US Army used machine learning via the Army Vantage Platform to identify $2.2 billion in unspent funds, and it found nearly $1 billion annually over the last two years. The military also faces compliance hurdles, as the GSA terminated its OneGov deal with Anthropic after the Pentagon designated the company a supply-chain risk. This designation followed months of disagreement over safeguards that the Defense Department argued were excessive. The GSA will now require AI companies to grant the government an irrevocable license to use their systems for legal purposes. The GSA also drafted rules requiring contractors to ensure "partisan or ideological judgments" are not embedded into data outputs. OpenAI also secured a $200 million contract to develop prototype frontier AI capabilities for the Department of Defense. This company aims to increase sales from $10 billion to $125 billion over four years, even though it lost $5 billion in 2024 on $4 billion in revenue. OpenAI also works to combine GPT-4o models with Anduril Industries’ systems to improve defenses against unpiloted aerial attacks. Will the government continue to grant large contracts to firms that cannot meet these emerging regulatory and supply-chain standards?