Japanese conglomerate SoftBank Group has completed its USD 30 billion investment in ChatGPT maker OpenAI as part of its commitment to the frontier AI lab’s last fundraising round.
Earlier this year, OpenAI had secured USD 122 billion in commitments that valued the Sam Altman-led venture at USD 852 billion. Amazon, Nvidia, and SoftBank had anchored the fundraising.
The Japanese investment conglomerate, led by founder Masayoshi Son, made public its commitment months ago to invest the capital through three tranches of USD 10 billion each into OpenAI through “Vision Fund 2.”
SoftBank, in September, raised USD 11.1 billion in the largest high-yield corporate bond sale globally to fund its bet on OpenAI.
SoftBank’s cumulative investment in OpenAI now stands at USD 64.6 billion, giving it a 13% stake in the AI lab.
OpenAI, in the coming months, is looking to raise at least USD 30 billion, seeking a valuation of about USD 1.4 trillion.
The fundraising discussions are at an early stage, and terms could change, reported Bloomberg, adding that the demand for the round is being led by investors.
The proposed round would serve as a bridge financing to provide capital in place of an IPO, with the latter being deferred over the safety concerns among the public and policymakers about the products developed by the frontier AI labs, including OpenAI.
The company had closed a funding round in March with USD 122 billion in committed capital, valuing it at USD 852 billion.
The Frontier AI lab’s annualised revenue run rate is nearing USD 70 billion, having risen by more than 70% since the beginning of the third quarter.
OpenAI’s rival, Anthropic, meanwhile, has reported a net loss of USD 42 billion, which the company witnessed in 2025.
The Dario Amodei-led venture is planning to spend USD 518 billion on cloud, computing, and infrastructure obligations.
While OpenAI has postponed its IPO, Anthropic has taken a different approach by highlighting the risks associated with its AI models in its IPO prospectus, stating that these models could exhibit “self-preserving behaviors” like attempts to “resist shutdown,” “conceal or manipulate information,” and behaviors “resembling blackmail.”