SoftBank was reported on January 29, 2025, to be discussing a direct investment of roughly $15 billion to $25 billion in OpenAI. It was not a completed $25 billion transaction. The proposal appeared alongside a separate approximately $15 billion SoftBank commitment to Stargate, potentially putting the group’s OpenAI-related spending above $40 billion. The talks later evolved into a formal, staged financing relationship that was larger than the original headline.
What the January 2025 report actually said
TechCrunch reported that SoftBank was considering investing up to $25 billion directly in OpenAI, with the discussed range approximately $15 billion to $25 billion. The terms were not finalized, so “SoftBank invests $25 billion” was not an accurate description at that point. If completed as reported, SoftBank could have become OpenAI’s largest single financial backer, depending on how “largest” was measured—by one investment, cumulative capital, ownership or voting power.
The report also described a separate SoftBank commitment of about $15 billion to Stargate. Combining the proposed direct investment and infrastructure commitment could have put SoftBank’s broader OpenAI-related exposure above $40 billion. The January coverage cited an OpenAI valuation of approximately $157 billion, a historical figure from late 2024 rather than a current valuation.
TechCrunch’s January 29 report also said OpenAI was negotiating a for-profit restructuring intended to facilitate additional fundraising.
#1 Best Overall
Why DeepSeek changed the context
DeepSeek’s R1 model had just attracted intense attention because investors interpreted its reported performance and comparatively low development-cost claims as evidence that capable AI might require less expensive computing than previously assumed. Nvidia lost as much as approximately $589 billion in market value in one day as markets questioned whether AI companies would continue buying vast quantities of advanced hardware, according to the same TechCrunch report.
That reaction raised three separate questions:
- Model efficiency: Can similar capabilities be achieved with less training compute?
- Inference demand: Even efficient models may create enormous usage and serving requirements as more people and businesses adopt them.
- Infrastructure strategy: OpenAI and its partners were planning capacity for reliability, deployment and future scale—not only for one model-training run.
DeepSeek therefore challenged the economics of an AI infrastructure build-out, but it did not establish that large data centers, chips or power projects were unnecessary. Lower training costs could reduce one category of spending while increasing experimentation, applications and inference demand.
Rank #2
OpenAI’s allegation about model distillation
OpenAI said it had evidence that DeepSeek used outputs from OpenAI models to train competing systems through “distillation.” OpenAI characterized that as a potential violation of its terms of service, which prohibited using outputs to develop competing models. This was an allegation by OpenAI, not an independently established legal or technical finding in the January report.
Stargate was a separate infrastructure project
Announced on January 21, 2025, Stargate was described as a new company intended to invest up to $500 billion over four years in U.S. AI infrastructure for OpenAI. SoftBank and OpenAI were lead partners: SoftBank was assigned financial responsibility and OpenAI operational responsibility. Oracle, MGX and other technology partners were also named.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchThe $500 billion was an intended multiyear infrastructure commitment, not $500 billion of immediately available cash and not a single equity check from SoftBank. The project was designed to finance data centers, electricity, chips and related facilities. Project equity, debt and partner funding could all form part of that total.
That makes Stargate different from a direct investment in OpenAI’s corporate balance sheet. The two initiatives were strategically connected—both supported OpenAI’s expansion—but they were not the same transaction.
Why OpenAI and SoftBank wanted the relationship
OpenAI’s objectives
- More capital for compute: Training and serving frontier models require expensive, long-lived infrastructure.
- Infrastructure diversification: A SoftBank-backed build-out could broaden OpenAI’s access to capacity rather than leaving it dependent on one primary cloud relationship.
- Restructuring support: A conventional for-profit structure could make large outside financing easier to arrange.
- Scale and reliability: Dedicated facilities could support growing demand and reduce capacity bottlenecks.
Diversification did not mean Microsoft disappeared. It meant OpenAI was seeking additional infrastructure and financing relationships.
SoftBank’s objectives
For SoftBank, the proposal offered strategic exposure to a leading model developer and a way to finance the infrastructure surrounding that developer. Masayoshi Son’s broader AI strategy depends on the possibility that model capability, applications and the physical systems that run them will become a large economic platform.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Best Value
That strategy also concentrates risk. A large investment only creates value if OpenAI can convert technical capability and infrastructure scale into durable revenue, margins and strategic returns.
The central investment case: bullish and bearish readings
| Bullish interpretation | Bearish interpretation |
|---|---|
| More efficient models could make AI affordable to more users, expanding total inference and application demand. | DeepSeek challenged the assumption that more spending and more chips automatically create a durable advantage. |
| SoftBank supplies capital, infrastructure expertise and a strategic partner willing to fund long-horizon projects. | Model capabilities may commoditize quickly, leaving expensive facilities with weaker pricing power. |
| Stargate could give OpenAI dedicated capacity and reduce bottlenecks. | OpenAI’s valuation, restructuring and infrastructure plans create execution and governance risk. |
| Scale may improve reliability and support large enterprise workloads. | SoftBank’s concentration increases financing risk if AI demand or asset values fall. |
The key test is not whether DeepSeek made AI cheaper in one dimension. It is whether lower costs increase total usage enough to justify the capital committed to capacity, power and hardware.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What happened after the headline
| Date | Development | What it means |
|---|---|---|
| January 21, 2025 | Stargate announced with an intended $500 billion, four-year U.S. infrastructure commitment. | A project-level infrastructure plan, not a $500 billion SoftBank equity investment. |
| January 29, 2025 | SoftBank reported to be considering $15 billion–$25 billion directly in OpenAI. | Negotiations, not a completed transaction. |
| March 31, 2025 | SoftBank and OpenAI entered a definitive agreement for up to $40 billion, subject to conditions. | The formal 2025 structure exceeded the original reported ceiling. |
| April 2025 | The first closing totaled $10 billion, including $7.5 billion from SoftBank Vision Fund 2 and syndicated participation. | Funding arrived in stages rather than as one immediate payment. |
| December 2025 | SoftBank completed an additional $22.5 billion investment. | The 2025 commitment was substantially funded through later closings. |
| February 27, 2026 | SoftBank announced a further $30 billion follow-on investment. | SoftBank said cumulative investment could reach $64.6 billion and approximately 13% ownership after completion. |
| April 1 and July 1, 2026 | Two $10 billion tranches of the 2026 follow-on were executed. | Official announcements confirm $20 billion of that new commitment; the retrieved material does not establish a third $10 billion tranche by August 18, 2026. |
The formal 2025 terms are described in SoftBank’s April 1 announcement; the December completion is documented here. SoftBank’s 2026 announcements covering the new commitment and the confirmed tranches are available at February 27, April 1 and July 1.
What investors should watch
- Funding mechanics: Announced commitments may involve syndication, conditions, debt and staged closings rather than immediate cash.
- OpenAI economics: Capital does not prove that revenue, margins and returns will support the valuation.
- Infrastructure utilization: Data centers and chips increase capacity but do not guarantee superior models or profitable demand.
- Governance: A transition from nonprofit roots to a heavily capitalized commercial structure can change accountability and control.
- Concentration: SoftBank’s exposure depends on OpenAI’s restructuring, valuation and ability to execute Stargate.
- Ownership versus influence: A projected 13% ownership figure is not the same as control, voting power or the largest cumulative economic stake.
Bottom line
The January 2025 story was a report about negotiations for up to $25 billion, not proof that SoftBank had already invested that amount. DeepSeek made the timing significant by forcing markets to question how much compute frontier AI really needs. It did not, by itself, invalidate OpenAI’s infrastructure strategy. The proposal instead became the opening chapter of a larger, conditional and staged financing relationship: up to $40 billion in the 2025 agreement, $22.5 billion completed in December 2025, and a further $30 billion announced in 2026 with two $10 billion tranches confirmed by July. The enduring question is whether OpenAI can turn that scale into sustainable economics before model efficiency and competition erode the returns on the infrastructure being built.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
