On Monday, May 4, 2026, the two biggest names in enterprise AI made the exact same strategic announcement on the exact same day. OpenAI launched a $10 billion joint venture called The Development Company, backed by TPG, Brookfield, Advent, and Bain Capital. Hours later, Anthropic announced a $1.5 billion joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs. Both ventures have one goal: deploy forward-deployed engineers to implement AI for mid-sized enterprises. Both ventures abandon the API-only model that defined the industry for the past three years.
This isn't a coincidence. This is a confession.
What OpenAI Announced
OpenAI's The Development Company raised $4 billion from 19 investors against a $10 billion valuation. Named investors include TPG, Brookfield Asset Management, Advent, and Bain Capital, with no overlap with Anthropic's investor list.
The venture will embed OpenAI engineers directly into portfolio companies owned by these private equity firms. Those companies get preferred access to OpenAI's engineering resources. The investors capture more value from AI contracts with their own companies. OpenAI gets guaranteed enterprise customers and consulting revenue that doesn't depend on per-token API pricing.
The model: Forward-deployed engineers (FDEs) who sit with customers, build end-to-end workflows, and take them to production. The same model Palantir pioneered and used to justify a $90 billion market cap.
What this tells CIOs and CTOs: OpenAI believes its API alone isn't enough to drive enterprise adoption at the scale they need for their $852 billion valuation (announced March 2026). They need bodies on-site.
What Anthropic Announced
Anthropic's joint venture is valued at $1.5 billion and includes $300 million commitments from Anthropic, Blackstone, and Hellman & Friedman. Additional backers include Apollo Global Management, General Atlantic, GIC, Leonard Green, and Sequoia Capital.
The announcement came from Anthropic's news page and was first reported by The Wall Street Journal. Like OpenAI's venture, Anthropic's new company will deploy engineers to mid-sized enterprises to build custom AI solutions. The venture gets preferred sales access to investors' portfolio companies.
Anthropic describes the model explicitly: "An engagement might begin with the company's engineering team sitting down with clinicians and IT staff to build tools that fit into the workflows that staff already use."
What this tells CFOs: Anthropic isn't selling you Claude API access anymore. They're selling you consulting engagements that look a lot like what Deloitte and Accenture sell, except with better AI models.
Why Both Ventures Launched the Same Day
This wasn't planned coordination. This was competitive pressure.
Bloomberg reported OpenAI's venture plans on the morning of May 4. Anthropic announced theirs hours later. Both had been in talks with private equity firms for weeks. Both saw the same strategic gap: APIs don't close enterprise deals fast enough.
The gap both vendors saw:
- Mid-sized enterprises don't have AI engineering teams
- Most companies can't integrate LLMs without serious help
- Professional services firms (Deloitte, Accenture, Capgemini) are too slow and too expensive
- Enterprise buyers want proof of ROI before committing to multi-year API contracts
- Investors (PE firms, VCs) want guaranteed deployment at their portfolio companies
The Palantir playbook both copied:
- Embed your own engineers into customer environments
- Build on your own stack (no third-party dependencies)
- Take solutions to production yourself (don't rely on customer IT)
- Charge consulting rates, not API rates
- Prove ROI before scaling usage
Palantir's stock is up 400% over the past two years using this exact model. Both OpenAI and Anthropic noticed.
What Forward-Deployed Engineers Actually Do
The FDE model isn't new. Palantir has run this playbook for 20 years. Here's what it looks like in practice:
Step 1: Embed engineers on-site. FDEs work directly with clinicians, finance teams, operations managers—whoever will use the AI tools. They don't work remotely. They sit in your building.
Step 2: Build workflows that fit existing processes. FDEs don't force you to rewrite your entire stack. They build AI agents and automations that plug into the tools your teams already use (ERP, CRM, EMR, SCM).
Step 3: Take solutions to production. FDEs don't hand off half-finished demos to your IT team. They own deployment, monitoring, and maintenance. If it breaks, they fix it.
Step 4: Scale usage incrementally. FDEs start small (one department, one use case) and expand as ROI proves out. No $10 million platform commitments on day one.
Why this model works for enterprise buyers:
- Faster time-to-value (weeks, not quarters)
- Lower risk (you're paying for outcomes, not API credits)
- No AI talent required (the vendor brings the engineers)
- Proof before scale (pilot with 10 users, expand to 10,000 if it works)
Why this model works for AI vendors:
- Higher revenue per customer (consulting rates beat API margins)
- Guaranteed adoption (PE-backed portfolio companies have to buy)
- Competitive moat (hard to replicate without talent density)
- IPO-ready metrics (recurring services revenue is more predictable than API usage)
What This Means for Enterprise Buyers
If you're a CIO or CTO evaluating OpenAI or Anthropic:
Option 1 (old model): Buy API access, hire your own AI engineers, integrate Claude or GPT into your stack yourself, hope your team can productionize it.
Option 2 (new model): Buy a services engagement, get vendor-supplied FDEs who sit on-site, let them build and deploy solutions for you, scale usage as ROI proves out.
The trade-off: Option 2 costs more upfront (consulting rates are higher than API rates), but it de-risks adoption. You're paying for guaranteed outcomes, not metered API usage that may or may not deliver value.
The catch: Both ventures prioritize their investors' portfolio companies. If you're not backed by TPG, Brookfield, Blackstone, or H&F, you're not getting first-tier FDE access. You'll get whatever engineering capacity is left over after the preferred customers get served.
What this tells you about vendor strategy: OpenAI and Anthropic both see services revenue as critical to hitting their $852 billion and $900 billion valuations. They can't get there on API usage alone. They need recurring, high-margin consulting contracts that look more like Palantir than AWS.
What This Means for CFOs
If you're a CFO evaluating AI spend:
The old AI budget model: Per-token API pricing. Pay for what you use. Scale usage as value proves out. Predictable unit economics.
The new AI budget model: Consulting engagements. Pay for FDE time (likely $300-500/hour loaded rates). Fixed-scope pilots that expand into multi-year contracts. Less predictable, but faster ROI.
The financial trade-off:
- API model: Lower upfront cost, slower time-to-value, higher risk of failed pilots
- Services model: Higher upfront cost, faster time-to-value, vendor owns deployment risk
The strategic question: Do you want to rent API access and build your own AI team, or do you want to rent vendor engineers who guarantee production deployments?
Most mid-sized enterprises will choose Option 2. They don't have the talent to hire AI engineers, and they don't have time to wait 18 months for a pilot to prove out.
The investor angle: If your company is backed by any of the 19+ investors in these ventures, expect sales calls. The PE firms didn't invest $4 billion and $1.5 billion without expecting guaranteed deployment at their portfolio companies. Your board may push you toward these vendors whether you want them or not.
Why This Shift Matters for the Industry
Three years ago, the AI vendor pitch was:
- "Here's an API. Integrate it yourself. Pay per token."
Today, the AI vendor pitch is:
- "Here's a team of engineers. We'll integrate it for you. Pay for outcomes."
What changed:
- Enterprises aren't adopting fast enough. API-only models depend on customers having AI engineering talent. Most don't.
- Professional services firms are too slow. Deloitte and Accenture charge $500/hour and take 12 months to deploy anything. Enterprises want faster results.
- Investors demand guaranteed deployment. PE firms didn't invest billions in AI vendors to watch their portfolio companies struggle with API integration.
- Valuations require recurring revenue. OpenAI's $852B valuation and Anthropic's $900B valuation require predictable, high-margin revenue streams. Consulting services fit that model better than metered API usage.
The bigger shift: AI vendors are becoming consulting firms. They're hiring FDEs, embedding them on-site, and building custom solutions. They're competing with Deloitte, not just with each other.
What this means for incumbent consulting firms: They're about to get squeezed. AI vendors have better models, better talent, and better unit economics. Consulting firms can't compete on speed or technical depth. They'll either partner with OpenAI/Anthropic or lose enterprise AI deals entirely.
What to Watch Next
If this model works, expect:
- Google to launch a similar venture (they can't let OpenAI and Anthropic own the FDE market)
- Microsoft to double down on Azure Consulting Services (they already have a services arm, but they'll need to hire faster)
- Meta to stay out (they don't have a cloud business, so they can't sell services at scale)
- Cohere and Mistral to launch smaller ventures (they'll need FDE capacity to compete for mid-market deals)
If this model fails, expect:
- OpenAI and Anthropic to quietly wind down the ventures after 18-24 months
- Investors to take losses (but they'll write it off as portfolio diversification)
- API-only pricing to come back as the default enterprise model
The key metric to track: How many of these FDE deployments actually go to production? Both ventures are betting that embedded engineers can close enterprise deals faster than APIs alone. If 70%+ of FDE engagements convert to multi-year contracts, this model wins. If conversion rates stay below 40%, this was an expensive experiment.
The timeline: Expect first results by Q3 2026. Both ventures will pilot 20-30 engagements over the next 6 months. If those pilots hit their ROI targets, the ventures will scale. If they don't, the investors will cut their losses.
The Bottom Line
On May 4, 2026, OpenAI and Anthropic both admitted the same thing: APIs aren't enough to drive enterprise adoption at the scale their valuations require. They need bodies on-site. They need guaranteed customers. They need consulting revenue that doesn't depend on per-token usage.
For enterprise buyers, this creates a choice:
- Rent API access and hire your own AI team
- Rent vendor engineers and let them deploy solutions for you
For most mid-sized enterprises, Option 2 wins. They don't have AI talent. They don't have time to wait 18 months for pilots to prove out. They need production deployments in weeks, not quarters.
For AI vendors, this shift is strategic survival. OpenAI can't justify an $852 billion valuation on API revenue alone. Anthropic can't justify $900 billion without recurring services contracts. Both need high-margin consulting revenue to hit IPO targets.
The confession both vendors made today: Selling AI models isn't enough. You have to deploy them too.
Want to calculate your own AI ROI? Try our AI ROI Calculator — takes 60 seconds and shows projected savings, payback period, and 3-year ROI.
Continue Reading
- Meta's $145B AI Bet Rejected, Google's $190B Rewarded - Why investors reward revenue proof over vision
- SAP Locks AI Agents Out. Salesforce Opens Every API - Competing enterprise AI integration strategies
- VC-Backed Firms Hit 80% AI Adoption: The Funding Signal - How investor pressure drives AI deployment
About THE DAILY BRIEF: Enterprise AI insights for technical and business leaders. Written by Rajesh Beri, Head of AI Engineering at a Fortune 500 security company.
