OpenAI has reached a post-money valuation of $300 billion following a $40 billion funding round led by SoftBank, the largest ever for a private tech company. But instead of a traditional capital raise, the company structured the deal as a tender offer—providing liquidity to employees and early investors through the secondary sale of existing shares.
Thrive Capital facilitated the transaction, drawing in strong demand from institutional backers aligned with OpenAI’s long-term strategy.
The deal underscores OpenAI’s transition into an infrastructure-focused AI powerhouse. As it pushes deeper into chip design, cloud partnerships, and national AI policy alignment, the company is distancing itself from its original nonprofit roots and even its foundational ties to Microsoft.
Shifting from Azure to Stargate Infrastructure
OpenAI’s realignment accelerated with its move away from exclusive reliance on Microsoft Azure. By February 2025, the company had started migrating workloads to infrastructure partners tied to the U.S.-backed Stargate Project, which aims to build $500 billion worth of domestic AI datacenters, chip fabs, and power systems over the next four years. The first $100 billion phase, focused on Texas, is already underway.
SoftBank’s investment has positioned the Japanese conglomerate as OpenAI’s largest investor, overtaking Microsoft. Of the $40 billion raised, SoftBank is contributing up to $30 billion directly and syndicating the remaining $10 billion to institutional partners, according to WSJ.
Microsoft still plays a role—its Copilot tools are deeply integrated with OpenAI’s models—but its influence is limited by evolving infrastructure agreements. Microsoft CEO Satya Nadella previously emphasized that “OpenAI APIs are exclusive to Azure going forward even, so nothing changes there… OpenAI has committed to Azure in a very significant way.”
Massive Compute and Power Needs Drive Strategic Investment
With AI models growing in scale and complexity, OpenAI’s infrastructure needs have skyrocketed. The company lost $5 billion in 2024 and expects compute expenses to exceed $9.5 billion annually by 2026. To help manage that, OpenAI signed a $11.9 billion, five-year compute deal with CoreWeave in March, which includes a $350 million equity stake.
This shift followed Microsoft’s retreat from its own CoreWeave contracts, indicating Azure’s AI expansion slowdown. By securing long-term GPU access through CoreWeave, OpenAI has avoided bottlenecks associated with Nvidia’s overstretched supply chain. That said, Nvidia’s Blackwell B200 GPUs—which OpenAI relies on—faced heat-related reliability issues last year, raising questions about sustained performance.
OpenAI’s chip roadmap now includes custom silicon. In partnership with Broadcom and TSMC, the company is developing dedicated AI chips to reduce dependence on Nvidia’s expensive GPUs. SoftBank’s ownership of Arm Holdings further bolsters these ambitions.
On the energy front, OpenAI is investing in Helion Energy, a U.S. fusion startup. Although commercial fusion power remains experimental, it represents a long-term bet on powering massive AI workloads sustainably. Microsoft has already signed agreements to purchase electricity from Helion, whose first power plant is projected for 2028.
Court Clears Legal Path for Tender Offer
The structure of OpenAI’s $40 billion funding round—a secondary share sale rather than a new capital raise—hinged on a March court ruling. Elon Musk had aimed to block OpenAI’s transition to a Public Benefit Corporation (PBC) in an ongoing lawsuit, arguing it violated the company’s founding charter. But the court rejected that argument on March 5, ruling that “there were no legal grounds to block the transition.”
This decision allowed OpenAI to proceed with its new structure, which lets it pursue commercial returns while retaining a nonprofit board and public benefit charter. Musk’s broader lawsuit, now fast-tracked, remains ongoing and includes claims about Microsoft’s influence over OpenAI’s direction.
Washington Strategy and Open-Weight Model Announcement
OpenAI is also aligning itself more closely with U.S. political priorities. The company has actively lobbied the Trump administration for regulatory clarity and economic incentives tied to AI infrastructure. Chief Global Affairs Officer Chris Lehane met with federal officials in March to advocate for consistent national policies.
OpenAI has proposed the creation of “AI Economic Zones” to accelerate AI deployment in sectors like energy, logistics, and manufacturing. Texas is expected to play a central role in this rollout, in part due to its existing alignment with the Stargate Project’s first infrastructure investments.
In a strategic move to counter competitive pressure from Meta and DeepSeek, OpenAI also confirmed it is preparing to release an open-weight AI model—the first since GPT-2. Open-weight models share pretrained parameters but withhold training data and source code.
This hybrid approach offers increased transparency while safeguarding proprietary techniques. The announcement is part of the company’s strategy to maintain competitive relevance against open-access rivals like China’s DeepSeek, which is expected to release its next-gen R2 reasoning model soon.
Tender Offer Rewards Insiders, Doesn’t Raise New Capital
Despite the headline-grabbing $300 billion valuation, the structure of this deal means OpenAI hasn’t taken in new operational funds. The tender offer is strictly a secondary transaction—no new shares were issued. It’s designed to reward long-term employees and early backers, while projecting confidence in OpenAI’s trajectory.
SoftBank’s role as the lead investor—contributing as much as $30 billion—marks a deepening of its AI ambitions and a strategic pivot away from earlier ties with Chinese models like DeepSeek. The Japanese conglomerate has since redirected its funding into OpenAI’s global rollout, including its newly formed SB OpenAI Japan unit.
Meanwhile, OpenAI continues to face unresolved legal and reputational risks. The company is still defending itself against a copyright infringement lawsuit from The New York Times, which alleges unauthorized use of its journalism in AI model training. That case, and broader questions about how OpenAI sources and manages its training data, could shape the regulatory climate around generative AI in the months ahead.
With a revamped governance structure, long-term infrastructure bets, and new model formats on the way, OpenAI is aiming to lead the AI sector moving forward. With competition become more intense, it remains unclear if it will succeed.
