Joint ventures. That’s the new weapon in the AI wars. OpenAI and Anthropic, locked in a fierce race for dominance, have teamed up with private equity giants to snap up engineering firms that make their models actually work in the real world. OpenAI’s new outfit, dubbed The Deployment Company, is deep into three deals. Expect an announcement any day now.
Five people close to the talks spilled the details to Reuters. The goal? Scoop up hundreds of engineers and consultants. Businesses crave these experts to tweak AI for their messy data, clunky systems, and shifting workflows. High-margin software dreams still hinge on sweaty, hands-on labor.
OpenAI pulled in $4 billion from 19 investors. Think TPG, Bain Capital, Brookfield Asset Management. Most of that cash targets acquisitions in the splintered world of IT services and consulting. Anthropic isn’t far behind. It’s chasing $1.5 billion from Blackstone, Hellman & Friedman, and Goldman Sachs, as The Wall Street Journal first noted.
Jon Gray, Blackstone’s president, couldn’t resist chiming in. “Hiring highly skilled workers will break down one of the most significant bottlenecks to enterprise AI adoption,” he said. His firm sees the joint venture expanding implementation partners. Smart. It echoes Palantir’s tactic: embed engineers right inside customer ops, tweak on the fly.
Both companies stayed mum. But the moves signal a pivot. Model-building grabs headlines. Deployment wins markets. Fragmented services firms are ripe for the taking. Consolidation ahead.
And it’s not just services. Hardware hunger rages on. Anthropic just kicked off early talks to buy inference chips from Fractile, a London startup. These DRAM-free accelerators promise lower costs amid shortages. Fractile’s SRAM design slashes pricey memory needs. A fourth supplier, joining Nvidia, Google, and Amazon. Chips roll out in 2027, per The Information and Tom’s Hardware.
Why diversify? Nvidia’s grip tightens. But cracks show. OpenAI tapped Broadcom for custom silicon last year. Anthropic weighs its own chips, sources told Reuters in April. Compute costs devour margins as inference surges past training expenses.
Nvidia CEO Jensen Huang regrets skipping early bets on these labs. Now, his firm pumps $30 billion into OpenAI, $10 billion into Anthropic. “Might be the last time,” he hinted at a Morgan Stanley conference, eyeing their IPO paths, according to CNBC. Vendor financing loops: Nvidia cash buys Nvidia chips.
Scale tips the balance. Anthropic locked 9.7 gigawatts across providers—AWS Trainium, Google TPUs, Nvidia GPUs. OpenAI trails at 5 gigawatts, mostly leased. Ownership matters. Inference battleground.
Enterprise push. That’s the endgame. Consumer apps? Free riders abound. But corporations pay for tailored AI. PE backers open doors to portfolio giants. Blackstone alone holds 250 companies. Instant sales channel.
OpenAI scrambles to match. Reports swirl of its own PE play. Late to the party. Anthropic’s run-rate revenue jumped from $9 billion to $30 billion in months. Cheaper inference via Fractile. Locked silicon stacks.
Risk lingers. Talent wars. Regulatory eyes. But deployment arms change everything. Models alone don’t cut it. Execution does.
Watch these ventures closely. They could reshape who owns enterprise AI. Not the builders. The deployers.
WebProNews is an iEntry Publication