Companies buying business data for AI training have turned private operational records into one of the most valuable assets a small or mid-sized business can hold in 2026. When Spirit Airlines sold a reported $12 million data licensing deal to an AI lab, the story broke across Fortune and Business Insider, and suddenly every CFO with a few hundred thousand records in their CRM started asking the same question: what is my data actually worth?
The answer, for many businesses, is far more than they expected. Deal sizes ranging from $100,000 to over $2 million are being reported across industries that have nothing to do with technology. Law firms, logistics operators, insurance agencies, and healthcare billing groups are among the businesses quietly signing deals and collecting payouts that dwarf anything they would have earned from a traditional asset sale.
Why AI Labs Are Paying for Your Business Data Right Now
To understand the market, you need to understand what the major AI labs are building right now. The next wave of AI systems is not chatbots. It is agents: software that can log into your tools, navigate interfaces, and complete complex multi-step tasks without a human in the loop. Anthropic, OpenAI, and Google DeepMind have each announced or deployed computer-use agents in 2025 and 2026. These systems can browse the web, fill forms, operate software, and complete business workflows without supervision. Building them requires a specific type of training data that cannot be generated synthetically: real recordings of how expert humans navigate real software to complete real tasks.
The public internet training data supply has effectively peaked. Common Crawl, the nonprofit that archives web pages for AI training, has indexed over 250 billion web pages, but AI researchers widely acknowledge that the marginal value of scraping more generic public text is diminishing rapidly. Meanwhile, legal pressure is mounting: the New York Times lawsuit against OpenAI, filed in late 2023, and similar suits from authors, publishers, and media organizations, have created significant liability risk around continued use of scraped public content. Labs are actively looking to shift toward licensed private data to reduce legal exposure and fill capability gaps.
Private operational data fills the gap on both fronts. It is scarce (it lives behind logins), it is legally cleaner (it comes with explicit licensing agreements), and it reflects exactly the kind of real human decision-making that agentic AI systems need to learn from. A business that has been running for five years has generated millions of workflow actions across its software stack. Each one is a potential training sample for the next generation of AI workers.
What Makes Your Data Valuable
Scale: More Employees Means More Trajectories
The fundamental unit of value in this market is the trajectory: a complete recording of one employee completing one multi-step task across one or more software tools. A business with 50 employees who each complete 20 to 30 workflow tasks per day generates tens of thousands of individual trajectories per week. Over five years, that accumulates into a dataset of millions of records. Scale is not the only variable, but it is the most straightforward: larger teams generate more data, and more data commands higher offers. This is why deal sizes jump significantly at the 30-employee, 100-employee, and 250-employee thresholds.
Scarcity: Niche Verticals Command Premiums
AI labs do not pay equal prices for all industries. They pay based on how much they need what you have. Data from software engineering, general customer service, and e-commerce is already abundant in existing training sets. Data from specialty verticals like healthcare administration, commercial insurance underwriting, specialty logistics, boutique legal, and skilled trades is genuinely scarce. Labs will pay a premium specifically because they cannot get it elsewhere. Per reporting in The Information, HVAC companies with rich service dispatch and job costing records have received offers that would have seemed absurd for operational data just two years ago. The scarcity of the vertical is the multiplier.
Recency: The Last 2-3 Years Are Most Valuable
AI labs are not looking to understand how people used software in 2015. They need to understand how people navigate the current generation of SaaS tools: Salesforce, HubSpot, NetSuite, ServiceNow, modern EHR platforms, current carrier portals. Data from the last two to three years reflects the actual software interfaces and workflow patterns that AI agents will need to operate. Older data is not worthless, but it carries a lower multiplier in pricing assessments. If your team has been running on a consistent software stack for the last three or more years, that recency is a specific asset. Businesses that recently upgraded their software stack and have two to three years of clean workflow data on modern platforms are in a particularly strong position.
Every Major Buyer Ranked and Compared
The table below covers every significant company actively buying business data for AI training in 2026. We have ranked them by accessibility and deal quality for the typical small to mid-sized business. Detailed notes on each platform follow the table.
| Company | Type | Pay Range | Headcount Min | Exclusivity | Referral Program |
|---|---|---|---|---|---|
| Polyshares | Data licensing marketplace | $100K to $2M+ | No stated minimum | Non-exclusive | Yes |
| Mode.inc | Trajectory data buyer | $50K to $500K+ | 6+ staff | Non-exclusive | Yes |
| Mercor | AI training data platform | Varies by project | 30+ | Project-based | No |
| Scale AI | Enterprise data labeling | Enterprise contracts | 100+ | Often exclusive | No |
| Endless Data | Data sourcing platform | Not publicly stated | 50+ | Unknown | No |
| Troveo | Vertical data marketplace | $25K to $300K | 20+ | Non-exclusive | Limited |
| Labelbox Prism | Enterprise annotation platform | Custom pricing | 100+ | Exclusive options | No |
| DataSupply.ai | Data broker (partner network) | Not publicly stated | 30+ desk employees | Not published | Yes (multi-level) |
Not sure if you qualify? Use our free estimator below to get a deal range in under 60 seconds:
Data Licensing Value Estimator
Answer five quick questions to see what your business operational data could be worth to AI companies.
Estimated deal range for your business:
$–
Estimates only. Actual deals vary based on data quality, exclusivity terms, and buyer demand.
Estimates only. Data licensing deal values vary based on data quality, organizational depth, exclusivity terms, and current buyer demand. Use this as a directional range, not a guarantee.
Want this estimate emailed to you?
Enter your email and we will send your personalised data licensing estimate straight to your inbox.
Detailed Notes on Each Platform
Polyshares: Best for Most Businesses
Polyshares operates as a full-service data licensing marketplace, meaning they do not simply pass your information to a single buyer: they run a competitive process that puts your dataset in front of multiple AI labs simultaneously. This approach typically produces better pricing than going direct to a single buyer, because labs are bidding against each other rather than setting their own price unchallenged. Deal ranges run from $100,000 on the low end to over $2 million for large, rare-vertical datasets. There is no publicly stated headcount minimum, which makes Polyshares accessible to businesses that might not meet the thresholds required by enterprise-focused competitors. Deals are non-exclusive, meaning you retain the right to license to additional buyers under separate agreements. Payout terms run net 30 to 60 days after deal close. For most businesses with 20 or more employees and a meaningful software stack, Polyshares is the right starting point.
Mode.inc: Best for Smaller Teams
Mode.inc specializes specifically in operational trajectory data from small and mid-size businesses, and it is the most accessible platform in the market for smaller operations. The minimum headcount is 6 staff, which opens the door for boutique law firms, accounting practices, healthcare billing groups, and small professional services firms that generate rich workflow data despite running lean teams. Mode’s intake process focuses heavily on the quality and variety of software tools your team uses daily: a 10-person law firm using four or five specialized platforms can qualify for a meaningful deal even without large headcount. Pay ranges run from $50,000 to over $500,000 depending on the depth and rarity of the dataset. Mode also runs a referral program for consultants, business advisors, and others who refer qualifying companies.
Mercor: Project-Based Engagements
Mercor approaches the market differently from Polyshares and Mode. Rather than offering a single lump-sum license deal, Mercor structures engagements as ongoing data collection projects where your employees participate in structured recording sessions. Compensation varies by project scope and the specific data requirements of the AI lab involved. The minimum headcount threshold sits around 30 employees, and Mercor tends to focus on businesses in sectors like finance, healthcare, and logistics where multi-step workflows are complex and repeatable. The project-based model can work well for businesses that want ongoing revenue rather than a one-time payout, but it requires more operational involvement from your team.
Scale AI: Enterprise Only
Scale AI is the market leader in AI data infrastructure for large enterprises, but it is not the right fit for most small and mid-sized businesses. Their engagements are designed for companies with 100 or more employees, and deals are typically structured as enterprise contracts with exclusivity provisions that can lock your data to a single buyer. For businesses that meet the threshold, Scale can produce very large contract values, but the exclusivity requirement means you cannot simultaneously license to other buyers. Unless you have significant negotiating leverage, the non-exclusive platforms tend to produce better total economics for most businesses.
Troveo: Vertical Specialist
Troveo operates as a vertical-specific data marketplace with a particular focus on sectors like healthcare, legal, and specialty logistics. The minimum headcount of 20 employees and pay range of $25,000 to $300,000 positions Troveo between Mode and Polyshares in terms of scale. They are worth considering as a secondary platform if you are already submitting through Polyshares, particularly if your vertical is one they specialize in. Troveo’s referral program is limited compared to Mode and Polyshares, so it is less relevant for advisors and consultants.
DataSupply.ai: Active Recruiter, Proceed with Caution
DataSupply.ai has gained significant visibility in 2026 through an aggressive partner recruitment campaign, most notably a structured training program led by their partner network. The underlying premise is legitimate: they broker operational data deals between businesses and AI labs including Micro1, and the demand for private business data is real. The Spirit Airlines deal that Micro1 pursued in September 2026 confirms this market exists.
The concern with DataSupply.ai is structural rather than conceptual. Their partner program operates on a multi-level referral model: partners can only join through existing partners, and a significant part of the pitch is recruiting more partners rather than closing data deals. Pay ranges and deal terms are not published publicly, and there are no independently verified closed deals on record. The emphasis on urgency during recruitment calls (“data is depreciating as we speak”) is a pressure tactic common in network marketing programs.
If you have already been pitched DataSupply.ai and are evaluating it, the core question is whether your connection is primarily interested in your data deal or in recruiting you as a partner. If it is the latter, we recommend starting with Polyshares or Mode.inc instead, both of which offer transparent terms, no partner recruitment requirements, and direct deal structures.
If you want to explore DataSupply.ai directly: business owners can check if their company qualifies here. Connectors and sales reps looking to refer companies can join the DataSupply.ai partner network here. Go in with clear eyes on the structure.
Labelbox Prism: Enterprise Annotation
Labelbox Prism is an enterprise-grade data annotation and curation platform primarily designed for companies building AI systems rather than selling raw operational data. Their 100-employee minimum and custom pricing structure puts them out of reach for most small businesses. Labelbox is most relevant if you have significant internal AI development efforts and need a data management partner, rather than if you are simply looking to monetize existing operational records.
Endless Data: Limited Transparency
Endless Data has emerged as a player in the data sourcing space, but public information about their deal structures, pay ranges, and exclusivity terms is limited. Their stated minimum of 50 employees positions them for mid-market businesses. Until more deal activity and pricing data becomes publicly available, we recommend starting with Polyshares or Mode and considering Endless Data as a secondary option if those platforms do not produce acceptable offers.
Our Top Picks
Best for Most Businesses: Polyshares
Polyshares is our top recommendation for the majority of businesses considering a data licensing deal. There is no stated headcount minimum, which means even a team of 15 to 20 can get a valuation. The deals are non-exclusive, so you retain the right to license to multiple buyers. Payouts run on a net 30 to 60 schedule, which is faster than most enterprise contract cycles. The competitive bidding process typically produces better pricing than any single-buyer approach. If you want to find out what your company’s data is worth, start your Polyshares valuation here.
Best for Smaller Teams and Law Firms: Mode.inc
Mode.inc works with companies as small as 6 staff, which makes it the most accessible platform for boutique professional services firms, law offices, accounting practices, and similar operations that run lean but generate rich workflow data. Law firms in particular are a strong fit: the multi-step navigation involved in case management, billing, court filing, and client communication is exactly the kind of trajectory data labs are paying top dollar for. If you are under 30 staff but have a meaningful software stack and at least a few years of operational history, check your eligibility with Mode.inc.
How to Get Started
The process for monetizing your business data is simpler than most people expect. Pick a platform: for most businesses, start with Polyshares. If you have under 30 staff, also submit to Mode. Complete their intake form, which asks about headcount, software stack, industry, and years of operation. No data is shared at this stage. The platform scores your dataset and returns an estimated deal range. Review the contract carefully, paying attention to exclusivity terms, anonymization protocols, and payout schedule. Non-exclusive agreements are almost always the better option for your long-term positioning.
Understanding what your business assets are actually worth is a core part of intelligent business ownership. Data licensing is one of the newer categories, but the economics are real and the window for premium pricing is open now. For context on how to think about business assets more broadly, our guide to buying a business covers valuation fundamentals that apply equally when you are the seller. And for a deeper look at the specific data type that commands the highest prices, see our full explainer on what operational trajectory data is and why labs will pay millions for it.
If your business is in the evaluation stage and you want to understand the capital landscape beyond data licensing, the full guide to small business funding options covers alternative capital sources that may complement a data licensing deal.
Ready to Know Where You Stand?
The Business Journey dashboard maps your exact position across all 13 stages. Track your progress, unlock resources for each step, and build with a framework used by thousands of founders at Hustler's Library.
No credit card required · Takes 3 minutes · Personalized to your stage
