Breaking from the traditional software vendor model, OpenAI has taken an equity stake in Thrive Holdings, an incubation vehicle created by Thrive Capital. The deal will see the AI giant embed its own engineers directly into accounting and IT service firms to automate operations from the inside out.
Vertical integration marks a significant pivot for the company, which was recently valued at $500 billion. Facing projected operating losses of $74 billion by 2028, OpenAI is under pressure to demonstrate that its models can replace human labor in the real economy, rather than just assisting it.
The Pivot: From Vendor to Owner
Under the terms of the agreement, OpenAI is not merely a technology partner but an equity owner in Thrive Holdings. Distinct from Thrive Capital, the venture firm led by Joshua Kushner that has repeatedly backed OpenAI, Thrive Holdings is designed to acquire and consolidate legacy service providers.
Targeting industries that generate hundreds of billions in revenue yet remain dependent on manual labor, the strategy focuses initially on accounting and IT services. These sectors are characterized by high-volume, rules-based workflows that are ripe for automation.
Promo
Thrive Holdings outlined the specific inefficiencies driving this selection in the partnership details:
“We are excited to partner with OpenAI to accelerate our mission by bringing frontier AI into foundational parts of the economy. Holdings currently operates in the accounting and IT services sectors, vast industries that produce hundreds of billions of revenue each year but still run on workflows that have barely changed in decades.
[…] The partnership will bring together a unique, cross-functional team, comprising OpenAI’s leading research and applied AI teams working alongside engineers, operators, and industry experts at Holdings to deeply integrate AI into the businesses that we own and operate. OpenAI will become an owner in Holdings as a result of this partnership, which we believe will align incentives for long-term value creation.
We believe the most powerful applications of AI will emerge from creating tight feedback loops between research, product, and engineering teams, in lock step with domain experts within businesses. By training the most advanced models for specific tasks within our businesses, guided by both company-specific data and expert feedback, we believe we can continuously improve model capabilities and ultimately establish AI as an integral driver of long-term enterprise value.”
By targeting these specific verticals, the partners aim to bypass the integration challenges that often stall enterprise AI adoption. Rather than selling software to a third-party CEO and hoping for implementation, Thrive Holdings will own the companies, granting OpenAI direct access to re-engineer their workflows.
To execute this vision, OpenAI will deploy “embedded” teams comprising researchers, product managers, and engineers directly into the portfolio companies. These technical experts will work alongside domain practitioners to identify bottlenecks that standard large language models cannot solve out of the box.
Joshua Kushner, CEO of Thrive Holdings, framed the strategy in a statement on X as a structural inversion of the typical software sales model, noting that “historically, technology transformed industries from the outside in. We believe this paradigm shift will happen from the inside out as domain experts and practitioners use AI as a native tool to reshape their fields.”
Such an “inside-out” approach attempts to create a repeatable playbook for turning services into software. If successful in accounting and IT, the model could be exported to other labor-intensive industries like legal services or medical billing.
Brad Lightcap, OpenAI’s COO, positioned the initiative in the official announcement as a scalable template for future enterprise deployment, stating that “this partnership with Thrive Holdings is about demonstrating what’s possible when frontier AI research and deployment are rapidly deployed across entire organizations to revolutionize how businesses work and engage with customers.”
The Financial Pressure Cooker
Capturing value from the services layer comes as OpenAI faces intensifying financial scrutiny. Following the $6.6 billion share sale finalized in October 2025, the company cemented a valuation of $500 billion, making it the most valuable private tech company in the world.
However, that valuation is underpinned by significant capital expenditures. As the company scales its infrastructure to support models like GPT-5 and beyond, costs are spiraling. Internal projections revealed in November indicate that OpenAI faces projected financial losses of $74 billion by 2028.
Subscription revenue from APIs and ChatGPT may not be sufficient to offset these infrastructure costs. By entering the services market, OpenAI can potentially capture a larger share of the economic value it creates.
Automating a $500-per-hour audit process offers significantly higher margins than selling the tokens used to perform it.
Accelerating this transition is essential for satisfying the company’s demanding backers. SoftBank, which recently executed SoftBank’s capital reallocation by selling $5.8 billion in Nvidia stock, is banking on OpenAI to justify its strategic pivot toward AI infrastructure.
Validating the economic utility of AI is essential to sustain the “Stargate” infrastructure project, a multi-phase buildout expected to cost over $100 billion. Investors need proof that the large-scale data centers being built will generate commensurate economic returns.
Such rapid expansion differs notably from competitors like Anthropic. Anthropic, by contrast, is targeting profitability by 2028 through a more traditional enterprise vendor model, avoiding the capital-intensive entanglements of owning service firms.
Vertical Integration vs. Horizontal Sales
OpenAI’s entry into the services market creates a complex dynamic with its existing ecosystem. For years, the company has acted as a horizontal platform, selling models to partners like Microsoft, Salesforce, and Accenture, who then build solutions for clients.
By taking equity in Thrive Holdings, OpenAI is effectively competing with the system integrators and consultancy firms that drive its enterprise adoption. A firm like Deloitte or Infosys may view this move as their primary vendor becoming a direct rival in the race to automate professional services.
Thrive Holdings argues that this conflict is necessary to break through technical ceilings.
Access to proprietary data is the central thesis here. External vendors often lack the granular “ground truth” data required to train models for specific, high-stakes tasks. By owning the operating entity, OpenAI can access every email, invoice, and log file needed to fine-tune its systems.
A permanent capital structure allows for experimentation that public companies or VC-backed startups might find difficult. Unlike a standard venture fund looking for a 5-7 year exit, Thrive Holdings claims a permanent capital structure.
Kushner highlighted this structural advantage, explaining that “our long-term orientation to transforming companies is possible because our time horizon at Holdings is forever.”
Structurally, the shift represents a broader industry trend toward “Service-as-Software.” Instead of selling a seat license for accounting software, the goal is to sell the completed audit itself.
This transition moves the revenue model from software margins to services revenue, potentially unlocking trillions in value if labor costs can be effectively removed.
The Broader AI Economy
The deal serves as a microcosm of the “asset-heavy” shift occurring across the technology sector. Major players are moving away from light-touch software sales toward capital-intensive operations. Meta and Google are spending hundreds of billions on data centers, effectively becoming utilities.
OpenAI is taking this a step further by becoming a service operator. If the “inside-out” transformation fails, the company risks being left holding equity in low-margin, human-dependent service firms, further weighing down its balance sheet.
However, if the experiment succeeds, it could validate the thesis that AI is a deflationary force capable of replacing human labor at scale.
Independent analysts and competitors have not yet publicly commented on the deal, leaving the long-term viability of this ‘inside-out’ model unproven against established consultancy giants.
The outcome of this partnership will serve as an important signal for investors questioning whether the current AI capital expenditure boom will ever yield a return.