**McKinsey Invested $4 Billion in a Company That Competes With McKinsey — and OpenAI Is the Reason**

DeployCo: From Model Maker to Embedded Deployment Partner

On May 11, 2026, OpenAI launched the OpenAI Deployment Company (internally called DeployCo), a standalone, majority-owned subsidiary backed by more than $4 billion in initial capital. It is not a product launch or a partnership program. It is a new company, with its own cap table, its own workforce, and a mandate to put OpenAI's own engineers inside enterprise clients to build production AI systems.

The structure is deliberate. OpenAI controls DeployCo through super-voting shares, giving the parent board-level authority over strategy, while the subsidiary operates as a separate business unit. That separation matters: DeployCo is meant to develop the operating model, pace, and client focus that enterprise deployment demands without being pulled into the cadence of OpenAI's model research cycle. At the same time, it functions as an extension of OpenAI, keeping clients connected to the research and product teams shaping the next generation of frontier models.

The investor consortium behind DeployCo is unusually broad. TPG leads, with Advent International, Bain Capital, and Brookfield as co-lead founding partners. Goldman Sachs, SoftBank Corp., Warburg Pincus, B Capital, BBVA, Emergence Capital, Goanna, and WCAS round out the founding partners (19 firms in total, per OpenAI's announcement). The consulting side is equally heavyweight: Bain & Company, Capgemini, and McKinsey & Company are all co-investors. Together, these firms sponsor more than 2,000 businesses worldwide, giving DeployCo a built-in pipeline of potential clients before it has delivered a single engagement.

The most unusual names on the cap table are McKinsey and Capgemini. Both compete directly with DeployCo for enterprise AI transformation revenue. Their presence as capital partners creates an economic incentive not to undermine the subsidiary they partly own, a dynamic that will be tested as DeployCo scales into their traditional client relationships.

Why a subsidiary, not a partner network

OpenAI already had distribution. Over one million businesses use its products and APIs. ChatGPT Enterprise and the API are self-serve: clients buy access and build on their own. DeployCo exists because that model hits a ceiling inside large organizations. The models are capable. The deployment is the bottleneck.

Most enterprises that have tried ChatGPT Enterprise or the OpenAI API have found that getting frontier models to reliably do useful work inside complex, messy real-world systems requires engineering effort they do not have. DeployCo's answer is to skip the handoff. Instead of selling tools and walking away, OpenAI embeds its own engineers inside the client organization to do the integration work directly.

This is the Palantir playbook, adopted almost wholesale. A Forward Deployed Engineer is not a solutions architect writing a scoped proposal. An FDE sits with the client's operators, identifies where AI creates the most value, selects a small number of priority workflows, and builds production systems that connect OpenAI models to the client's data, controls, and business processes. The FDE stays until the system is live, monitored, and handed to an in-house team that can maintain it.

OpenAI Chief Revenue Officer Denise Dresser framed it directly: "AI is becoming capable of doing increasingly meaningful work inside organizations. The challenge now is helping companies integrate these systems into the infrastructure and workflows that power their businesses. DeployCo is designed to help organizations bridge that gap and turn AI capability into real operational impact."

The Tomoro acquisition: 150 engineers on day one

To staff DeployCo from launch, OpenAI agreed to acquire Tomoro, an Edinburgh and London-based AI consulting and engineering firm founded in 2023 in explicit alliance with OpenAI. The deal has not yet closed (it remains subject to regulatory approvals expected in the coming months), but the intent is clear: Tomoro's approximately 150 engineers become DeployCo's day-one delivery workforce.

Tomoro's entire staff are FDEs. They have never done traditional consulting. Their operating model is embedding inside client environments and building production AI systems, and their client roster reflects that focus: Tesco, Virgin Atlantic, Fidelity International, Red Bull, Mattel, Supercell, and the NBA.

The Supercell engagement is the most detailed case study available. Tomoro built an in-game support agent for the maker of Clash of Clans serving 110 million users across five games. It launched in 12 weeks, processes 500 million daily tokens on GPT-4o and 200 million on GPT-4o-mini, and cut per-ticket resolution cost by roughly 90%. Customer satisfaction scores rose by 20%. Average response time is seven seconds. These figures come from OpenAI and Tomoro and have not been independently verified, but they represent the kind of ROI claim that wins enterprise procurement decisions.

Before the acquisition, Tomoro grew monthly revenue tenfold in 12 months and quadrupled headcount. London remains its European hub; APAC operations run from Singapore with offices in Sydney and Melbourne. The Tomoro brand is expected to be absorbed into DeployCo within one to two years of closing.

DeployCo will deploy exclusively on OpenAI models. It will not deploy Claude or Gemini. That constraint is structural, not incidental. It is the reason OpenAI built a subsidiary rather than partnering with independent consulting firms that can recommend whichever model fits the client's needs. The trade-off is clear: DeployCo captures more of the total contract value per deployment than OpenAI would earn from API usage generated through a third-party engagement. The risk is that it must now build its own client relationships, hire and retain FDEs at scale, and compete with the very consulting firms that are its co-investors.

Inside the Forward Deployed Engineer Role

A Forward Deployed Engineer at OpenAI doesn't build internal products. The job is to embed inside a customer's organization, take an ambiguous enterprise problem, and ship a production system powered by frontier models, then make sure it stays running.

The role spans the full arc: technical discovery, scoping, system design, build, and production rollout. FDEs "own technical delivery across multiple deployments from first prototype to stable production," according to OpenAI's own job listings, writing production-grade code in Python or JavaScript across frontend and backend while partnering directly with customer engineering teams. They measure success not in features shipped but in production adoption, measurable workflow impact, and eval-driven feedback that changes OpenAI's model and product roadmaps.

That last piece is the part most people miss. An FDE doesn't hand off a deployment and move on. They channel field friction (edge cases, performance gaps, integration failures) back to OpenAI's Research and Product teams. The role is bidirectional by design: deploy, learn, feed back, repeat.

What separates it from solutions engineering or MLOps

A solutions engineer demonstrates. An MLOps engineer maintains. An FDE owns the outcome. The distinction shows up in the details. OpenAI's postings require five or more years of engineering experience, including customer-facing work, plus direct experience building or deploying systems powered by large language models. FDEs are expected to contribute code when progress depends on it, not delegate to a delivery team. They scope work, sequence delivery, remove blockers, and make trade-offs between scope, speed, and quality, all while embedded on-site with the customer up to 50% of the time.

The government-facing variant adds another layer. The FDE-Gov role, based in Washington DC, Seattle, or San Francisco, requires an active TS/SCI clearance and familiarity with cloud deployment models like Azure, AWS, Kubernetes, and Terraform. These engineers work with defense and intelligence stakeholders, where delivery is urgent and ambiguity is the default.

The skill profile

Full-stack coding proficiency is table stakes. The differentiator is what Paraform's June 2026 role breakdown calls "AI-native deployment skill": building RAG pipelines, designing evaluation frameworks, and running agent workflows in production. Problem decomposition runs through everything. FDEs take a vague enterprise need, break it into scoped technical workstreams, and deliver under shifting constraints.

The interview process reflects that. The decomposition round, where candidates face an ambiguous large-scale problem and must clarify scope before writing a single line of pseudocode, is where most people wash out. Interviewers want continuous narration of reasoning, not a polished answer delivered in silence. It mirrors the actual job more than any algorithmic coding challenge does.

Where the roles are concentrated

New York now accounts for 35% of all FDE postings, per Agile Leadership Day India's career guide, surpassing San Francisco at 11%. The NYC concentration tracks with fintech and compliance-heavy industries that need hands-on deployment support. OpenAI's own footprint stretches to Seattle, Washington DC, and international offices in London, Munich, Paris, Singapore, Tokyo, Sydney, and Dublin. Active FDE postings in Stockholm and Madrid signal the role's expansion into European markets.

Most positions follow a hybrid model: three days in-office, with travel layered on top. Fully remote FDE roles are rare. The core mechanic of embedding on-site with customers makes permanent remote work structurally incompatible with the job.

The compensation matches the demands. OpenAI's listed base range runs from $145,800 to $280,000 depending on the posting, with equity on top. Paraform's breakdown, citing GetPerspective's 2026 compensation report, puts total comp at $350,000 to $550,000 at mid-to-senior levels, with equity doing the heavy lifting at staff tiers. That sits 60% above traditional solutions engineering comp, a gap that reflects the AI-literacy premium and the competitive pressure among OpenAI, Anthropic, and Google to lock down deployment talent.

Why Stockholm Is Becoming OpenAI's European Deployment Anchor

OpenAI's Forward Deployed Engineer posting for Stockholm is not an outlier. It sits among 23 active FDE roles on the company's careers page spanning Dublin, London, Munich, Paris, Madrid, Singapore, Tokyo, and Sydney. But Stockholm's placement tells a specific story about where OpenAI is building its European deployment muscle, and why the Swedish capital is emerging as a key operational hub alongside London.

The Stockholm FDE role carries the same structural mandate as its counterparts elsewhere: own technical delivery from discovery through production rollout, embed with customer teams, and feed deployment friction back to the parent's Research and Product groups. The posting requires the same senior-level background: five-plus years of engineering, full-stack proficiency in Python or JavaScript, and direct exposure to LLM or generative-model systems in production. Travel up to 50% is expected. Three days per week in the office. Relocation assistance offered.

Those details matter less for what they say about one job than for what the cluster of European FDE postings signals about OpenAI's enterprise strategy. London and Dublin have deeper financial-services concentrations. Munich and Paris have larger local tech labor pools. But Stockholm sits at the intersection of two things OpenAI needs: a dense base of engineers who already work across European markets, and proximity to the Nordic and Baltic enterprise sector that has been faster than most of Europe at adopting API-driven AI workflows.

OpenAI's own hiring data supports the pattern. Zero G Talent's board shows 48 OpenAI roles added in the past seven days alone, with FDE openings in Stockholm and Madrid among the newest listings. The company is not testing European demand; it is staffing for it at scale.

The broader context makes the timing clear. OpenAI has published an EU Economic Blueprint with 20 proposals to accelerate AI adoption across the continent, timed just before the European Commission's Apply AI Strategy rollout. The company also launched a $150 million Partner Network to fund global integrators building enterprise AI deployments. Stockholm is where those policy and partnership bets translate into headcount.

London remains OpenAI's most visible European office. Tomoro's headquarters are there, and most coverage has focused on it as the company's regional anchor. But the FDE hiring map suggests a complementary operational logic. London handles the finance, policy, and acquisition-integration relationships. Stockholm, Dublin, and Munich handle the actual deployment work: engineers on-site with European enterprises, running the feedback loop that makes frontier models usable in production.

The McKinsey Co-Investment Signal and What It Reveals

The investor consortium backing DeployCo is where the structural signal gets interesting. Among the 19 firms in the cap table sit McKinsey & Company and Capgemini, two of the largest consulting and systems integration firms in the world, both of which directly compete with the work DeployCo is designed to do.

That is not a contradiction. It is a calculated alignment. McKinsey and Capgemini co-investing in a subsidiary that could cannibalize their own enterprise AI transformation revenue creates an economic incentive to cooperate rather than compete aggressively. Both firms likely concluded that if OpenAI was going to build a deployment consulting business regardless, it was better to have financial exposure to its success than to fight it from the outside with no upside. The co-investment also gives both firms early access to DeployCo's operational playbooks and client intelligence, which they can use to inform their own service offerings.

The private equity sponsors add another layer. TPG, Advent, Bain Capital, Brookfield, and the other PE backers collectively sponsor more than 2,000 businesses globally. That gives DeployCo a captive channel into portfolio companies already under pressure from their sponsors to boost productivity through AI. The consulting and systems integration partners, which work with thousands more, provide a broad view of where AI creates value and which deployment patterns scale across industries.

Together, the Tomoro workforce inheritance and the investor consortium reveal what OpenAI is actually building: not a consulting arm that advises and leaves, but a deployment engine that embeds engineers inside client organizations, builds production systems connected to OpenAI models, and captures the implementation revenue directly. The 150 Tomoro engineers are the proof of concept. The McKinsey co-investment is the signal that the major professional services firms see this model as a structural shift, not a side project.

Who OpenAI Is Hiring — and What It Pays

DeployCo's hiring pipeline is already visible on Zero G Talent's board, and the roles tell you exactly how OpenAI is staffing its deployment subsidiary.

In the past seven days alone, OpenAI added 48 roles to the board, a pace that signals DeployCo is building out fast, not running a pilot. The listings span multiple cities and map to a clear division of labor.

Stockholm and Madrid each show Forward Deployed Engineer openings (the core DeployCo role, engineers who sit inside client organizations and wire frontier models into live production systems). These are the people Tomoro's 150 inherited FDEs were doing, and OpenAI is now hiring more of them on the ground in Europe.

London has a Program Manager posting for "Technology Capital Builds," a role that sits at the intersection of DeployCo's deployment work and the infrastructure investment side of the business, coordinating the build-out of client-facing AI systems at scale.

San Francisco is where the platform and tooling roles cluster. The listings include a Full Stack Software Engineer for "Agent Enablement," a Data Engineer for "Scaling Analytics," and an Android Engineer on the Growth team. These aren't FDE roles; they're the internal tooling and infrastructure positions that support the engineers deployed in the field.

The compensation spread is wide by design. The $230,000 Android role and the $405,000 agent enablement role occupy very different parts of the stack, and OpenAI is paying accordingly. The San Francisco data engineer range sits at the top end of what most AI labs offer for infrastructure-adjacent roles, which suggests DeployCo is competing for the same talent pool as the core OpenAI research organization.

What's missing from the public listings is just as telling: there are no junior or mid-level FDE postings in Stockholm or Madrid. OpenAI is hiring senior engineers for the embedded deployment role from day one, people who can walk into a Fortune 500 environment and ship production AI without hand-holding. That's a deliberate staffing choice. It's also why the Tomoro acquisition mattered: 150 experienced FDEs gave OpenAI a running start that no hiring sprint could replicate.

The Stockholm and Madrid roles don't list compensation ranges on the board. That's typical for European AI roles at this level (companies often negotiate individually rather than posting bands), but it also means the real salary signal will come from what competitors like Anthropic are offering their own forward-deployed and solutions engineering teams in those markets.

The Ripple Effect on AI Talent Markets

DeployCo's launch doesn't just create jobs at OpenAI. It redraws the boundary of what an AI lab expects from its engineers, and every competitor with a frontier model will feel the ripple.

Until now, the talent war in AI has orbited a relatively clean split: researchers who build models, and application-layer engineers who wrap APIs into products. DeployCo's Forward Deployed Engineer role collapses that divide. An FDE is expected to understand OpenAI's model capabilities deeply enough to architect a production deployment, then stay inside the client's organization long enough to make it work, writing code, navigating legacy systems, and iterating against real business constraints. That is not consulting. It is not MLOps. It is a new category: the engineer who ships frontier AI into enterprise production, end to end.

The signal is already visible in hiring velocity. OpenAI added 48 roles to its careers page in the past seven days alone, spanning Stockholm, Madrid, San Francisco, and London. Several carry titles (Forward Deployed Engineer, Program Manager for Technology Capital Builds) that didn't exist on the board a year ago. Compensation bands for related engineering roles in San Francisco sit between $230,000 and $405,000 a year, per current Zero G Talent data, placing FDE-calibre candidates firmly in the senior-staff bracket. That pricing tells you how scarce this profile is and how much OpenAI is willing to pay to lock it down before Anthropic, Google, or Mistral build their own deployment armies.

The structural implication runs deeper than headcount. McKinsey's co-investment in DeployCo signals that the world's largest management consultancy sees embedded AI deployment as a service category worth billions, not a one-off engagement but a permanent layer between model makers and enterprise buyers. When McKinsey bets on a talent model, the rest of the consulting and systems-integration industry follows. That means the demand for engineers who can operate at the intersection of frontier models and enterprise IT isn't a temporary spike. It is the beginning of a permanent talent market.

For engineers reading this, the practical takeaway is straightforward: if you can build with large-language-model APIs and you're comfortable working inside a Fortune 500's technology stack, not alongside it but inside it, you are now one of the most hireable profiles in AI. DeployCo proved the category exists. The competition to staff it is just starting.

Working in AI? Zero G Talent tracks the openings: browse AI jobs, openings at OpenAI, and the people building the field.