Well, if anyone thought the software hype era was over, think again.
In case you missed it, on Wednesday, the Pentagon’s Chief Digital and Artificial Intelligence Office (CDAO) increased its enterprise agreement with AI integration and training giant Scale AI from $100M to $500M, supporting “the Department’s strategic imperative to accelerate the adoption of data, analytics, and AI capabilities across the joint force to maintain a competitive advantage.”
Under the agreement—an OTA that’s set to last five years, if we want to be technical—anyone across the Pentagon gets a “streamlined, turn-key pathway to access Scale’s end-to-end AI capabilities – spanning data labeling, model development, generative AI platforms, and engineering services.”
And it sounds like they’re already using it.
“We are seeing traction across all components of the Department—including CDAO, Office of the Secretary of War, Defense Agencies, the Military Services, and more,” Scale Head of Defense Kathryn Harris told Tectonic. “The use-cases span the Department’s highest priorities missions, including autonomy and homeland defense.”
“Our platform is designed to help the Department operationalize its Responsible AI principles, ensuring that as we scale these capabilities, they remain reliable, auditable, and trusted by the warfighter,” she added in a statement.
Nuts and bolts: Now, Scale AI isn’t an AI company in that they just build LLMs—they’re an AI company in that they also build all of the data and software infrastructure and training that makes those LLMs work.
Here’s how their tech works:
- They prepare data for AI by collecting, cleaning, labeling, and organizing massive datasets (text, images, video, sensor data) so machine learning models can actually learn from them.
- They also test AI models by providing tools to evaluate, benchmark, and fine-tune AI systems, including generative AI models.
- They also make AI models themselves, designed specifically for secure and sensitive environments.
- That’s particularly important for AI being used in sensitive environments—say, inside a certain five-sided building. Scale makes sure that the AI being used for things like intel analysis, planning, and autonomous systems is reliable, secure, and, like, works.
And to say the company has been a hit would be the understatement of the century.
- Their tech is used by a few companies you may have heard of: OpenAI, Meta, Microsoft, Google, and Nvidia, among them.
- Meta invested $15B (yes, with a B) in the company last year, equating to about 49 percent ownership of the company.
- On the defense side of things, they’ve scored multiple $100M+ contracts with the Pentagon, including a $250M ceiling BPA with the Joint Artificial Intelligence Center in 2022 and a $99M Army R&D contract last August, plus that $100M-now-$500M enterprise agreement with CDAO.
- Harris said that within the department, “Scale is partnered with every Military Service, the defense innovation ecosystem (including CDAO, DIU, and others), multiple Defense Agencies and multiple Combatant Commands.”
We’re all friends here: Scale AI basically helps the Pentagon and government integrate and operationalize AI models and solutions safely and securely.
“Take Golden Dome, where we are working with a diverse coalition of industry and government partners, from well-known defense tech companies to early-stage startups and traditional defense primes, to develop Golden Dome’s C2 system,” Harris said by way of example. “Scale is making sure the agentic layer of the system excels at transforming information into insights, and insights into decisive advantage.”
And now, under this new contract, they’ll be doing a whole lot more of all that integration and testing goodness.
Anyone across the Pentagon—including those who don’t have a contract with Scale AI—will now have access to Scale tech at a pre-negotiated bulk price, including:
- Scale Data Engine: Scale’s bread-and-butter machine learning operations platform “to build, test, and deploy computer vision models, powered by expert-labeled, AI-ready data.”
- Scale GenAI platform (SGP): A platform purpose-built for government developers to “fine-tune, test, evaluate, and deploy generative AI models safely and securely on classified networks.”
- Scale Donovan: A generative AI platform designed for “defense and intelligence operators to turn massive amounts of unstructured data into actionable insights and decision support at mission speed.”
- Engineering capability development sprints: Basically, bespoke software and data pushed to meet “specific mission requirements.”
Harris said that under the initial contracts, they’ve seen a few usage trends:
- Donovan has been super popular with combatant commands for processing battlefield intelligence.
- Data Engine has been used to generate “mission-specific, expert-validated data.”
- SGP has been used to build agentic solutions that automate complex workflows.
Harris sees Scale’s role—and these tools—as critical to making sure the Pentagon actually delivers on its promises of deploying AI at scale, and in the most sensitive of environments.
“As the era of small-scale AI pilots gives way to enterprise-wide deployment, the infrastructure decisions the Department makes known—which platforms to trust, which data operations to invest in, which partners to build for the long term—will shape the operational AI capabilities of the U.S. military for the decade ahead,” she said in a statement.