SoftBank’s Year-End Dash: Racing to Secure $22.5 Billion for OpenAI

It's a high-stakes sprint against the calendar. SoftBank is pushing to lock down the final piece of a massive financial commitment—a cool $22.5 billion—to its artificial intelligence partner before the year flips.
The Clock is Ticking
Forget quiet holiday planning. The investment giant's corridors are buzzing with activity, focused on one number: the remaining balance owed to OpenAI. This isn't about a simple transfer; it's a complex financing operation hitting its final, critical phase. The pressure to close this chapter by December 31st adds a layer of urgency that markets are watching closely.
Why the Rush?
Year-end deadlines in finance are rarely just administrative. They signal strategic positioning, tax considerations, and a desire to present a cleaned-up balance sheet to investors. Securing this tranche solidifies one of the most significant tech partnerships of the era and prevents any funding uncertainty from spilling into the new year. It’s a move to maintain momentum in an AI arms race where hesitation is not an option.
The Bigger Picture
This final push underscores a relentless bet on AI as the definitive frontier. For SoftBank, it’s about securing a pole position. For the broader market, it’s a bellwether of conviction—a signal that major capital continues to flow into foundational AI platforms, regardless of economic whispers. Yet, one can't help but muse on the sheer scale: $22.5 billion is a sum that could bail out a small country, but here it's just the remaining tab for a single tech bet. The future is expensive, and someone’s got to foot the bill.
SoftBank pulls cash from assets, loans, and delayed listings
To raise more money, SoftBank is preparing several routes at once. One of the biggest is the long‑planned listing of PayPay, its payments app unit.
The IPO was first expected this month but slipped because of the 43-day U.S. government shutdown, which ended in November. Sources now say the listing should land in the first quarter of next year and could raise more than $20 billion if market conditions cooperate.
The group is also looking to trim its position in Didi Global, China’s largest ride‑hailing company. Didi plans to list shares in Hong Kong after being forced to exit U.S. markets in 2021 following a regulatory crackdown. A source with direct knowledge allegedly said SoftBank is exploring exits tied to that move.
Beyond asset sales, Masa has plenty of financial tools ready. SoftBank expanded its margin loan capacity by $6.5 billion, lifting total unused borrowing power to $11.5 billion. Those loans are backed by its stake in Arm Holdings, whose stock has tripled since its IPO, giving SoftBank more collateral room.
As of September 30, SoftBank also reported 4.2 trillion yen, or $27.16 billion, in parent‑level cash. It still owns roughly 4% of T-Mobile US, a stake valued near $11 billion, according to LSEG data.
Despite cutting back overall activity, SoftBank has kept funding select AI startups, including Sierra and Skild AI, even as most capital flows toward OpenAI.
OpenAI pushes spending as compute demands explode
The money matters because OpenAI needs it fast. The company has not yet received the remaining funds but expects payment by the end of 2025, as outlined in its contract.
Both OpenAI and SoftBank back Stargate, a $500 billion effort to build massive AI data centers for training and inference, which executives say aligns with U.S. goals to stay ahead of China as Donald TRUMP sits in the White House in 2025.
Big tech firms are spending heavily on similar infrastructure. Meta Platforms and others are pouring cash into chips, power, cooling systems, and servers, often bringing in partners to spread risk.
That spending wave has raised worries about returns and the chance of an AI bubble if revenues fail to match costs.
SoftBank agreed in April to invest up to $30 billion in OpenAI. $10 billion landed immediately. The rest depended on OpenAI completing its shift to a for‑profit structure by year‑end, a change the company finalized in October.
Costs inside OpenAI keep climbing. Training and running models is getting more expensive as competition from Alphabet’s Google intensifies. Sam Altman recently told staff the company had entered a “code red” phase to upgrade ChatGPT, delaying other launches to counter momentum behind Gemini.
In October, Sam said OpenAI aims to build 30 gigawatts of compute for $1.4 trillion, with a long‑term goal of adding 1 gigawatt every week, a scale where each gigawatt now costs over $40 billion.
Get $50 free to trade crypto when you sign up to Bybit now