Nvidia Reported in Talks to Guarantee About $250 Billion of OpenAI Data-Center Financing; Shares and Credit Insurance Move Sharply
A Wall Street Journal report that Nvidia may backstop lease and debt payments for a 10-gigawatt OpenAI campus in Ohio built by SoftBank's SB Energy sent Nvidia stock down about 5% and pushed the cost of insuring its debt to its biggest one-day rise on record, renewing a dispute over whether AI industry deals are self-reinforcing.
A Record Jump, and Nothing Signed
On Monday, July 27, 2026, The Wall Street Journal reported that Nvidia is in talks to guarantee about $250 billion in financing for OpenAI[1][2]. Nvidia shares fell roughly 5% that day, to around $197[5]. The cost of insuring Nvidia's debt against default jumped 14 basis points in a single session, to 82 basis points — the sharpest one-day move since that market began trading actively[6].
Here is the tension underneath those numbers. Nothing has been signed. Reuters said it could not independently verify the report, and people familiar with the talks cautioned the deal could still fall apart[1]. Yet Nvidia's own bondholders reacted as if real money had already moved.
That gap — a market treating an unconfirmed term sheet as a near-certain event — is the story. To understand why credit investors reacted so much harder than anyone expected, you have to understand what Nvidia is actually being asked to promise, and why a promise that costs nothing today spooked people who insure debt for a living.
What Nvidia Would Actually Be On the Hook For
The project at the center of this is real and mostly public. OpenAI, SoftBank Group and SoftBank subsidiary SB Energy have announced a partnership to build a 10-gigawatt data-center campus in Pike County, Ohio, on federal land that once housed a uranium enrichment plant[8][9][10]. Ten gigawatts is roughly the output of ten large nuclear reactors. Powering it requires about 9.2 gigawatts of new natural gas generation and $4.2 billion of transmission work with the utility AEP Ohio, with an initial 800-megawatt phase expected by 2028[8][11].
The financing is the part still in talks. The Journal's report says Nvidia would guarantee lease and debt payments so OpenAI can occupy the campus, plus a separate discussion of up to $350 billion to finance OpenAI's purchases of Nvidia chips[1][3].
A guarantee is not a loan, and the difference matters. Nvidia would not hand OpenAI cash. It would promise lenders and the landlord that if OpenAI can't pay, Nvidia will[1]. That promise is what's called a contingent liability: it costs nothing up front and doesn't show up as debt on Nvidia's balance sheet. It only becomes real money if OpenAI actually defaults. Credit investors watch this kind of promise closely for exactly that reason — it hides real exposure in the footnotes rather than the debt total, which is part of why bond markets moved more sharply than the stock market did[6].
The guarantee matters because OpenAI doesn't have an investment-grade credit rating[7]. Analysts cited in coverage project OpenAI will lose roughly $14 billion in 2026 on about $25 billion in revenue[7]. Lenders won't extend cheap financing against numbers like that on their own. Attach Nvidia's much stronger credit to the deal, though, and the math changes — lenders price the debt off Nvidia's name instead of OpenAI's[1][7].
"Around and Around We Go"
That's the phrase investor Michael Burry used to summarize the core worry, and it names the fight in one line[15]. The argument: if Nvidia guarantees the rent on a building whose tenant's main expense is buying Nvidia chips, then some of what looks like independent demand for those chips is really Nvidia's own credit finding its way back to Nvidia's own revenue.
The skeptics' strongest evidence is historical. In the late 1990s, Lucent and Nortel financed telecom customers who couldn't get credit anywhere else. When those customers collapsed, Lucent had to reverse revenue it had already booked and wrote off billions[14]. Short sellers and credit analysts see the same shape now: a chipmaker extending its name to keep its own customers afloat. Their incentive is worth naming plainly — short sellers profit if AI valuations fall, and bond analysts are paid to price downside risk, so pessimism is built into their job[13].
Nvidia's counter isn't just "trust us." It rests on a structural distinction its defenders think critics are skipping past. Lucent booked the revenue immediately and carried the customer's IOU as a loan already made. Nvidia's exposure here is a contingent promise on a lease, not cash it has already lent out[13][14]. Nvidia has sent Wall Street analysts a memo stating it does not do vendor financing, and CEO Jensen Huang has argued its stakes in customers like CoreWeave are small next to the much larger sums those companies raise on their own[4][13].
Nvidia's underlying case is also about what's actually constraining its business. The bottleneck for AI isn't chip demand — it's buildings and power plants, which are slow and expensive to finance. Lending its credit rating costs Nvidia nothing unless OpenAI defaults, and it's a far cheaper way to unlock that construction than building the data centers itself[1].
The Company With the Most to Lose Twice
If there's a stakeholder whose position deserves a second look, it's SoftBank. From outside, it's easy to read the Ohio project as one more AI real estate story. Inside the structure, SoftBank is exposed twice over to the exact same bet.
SoftBank is both an equity investor in OpenAI — a stake partly funded by a $40 billion bridge loan now being syndicated to 21 additional banks[10][19] — and, through SB Energy, the developer building the campus OpenAI would lease[8][10]. If OpenAI stumbles, SoftBank takes the hit on the equity side and the landlord side at once. That's why Japanese markets, rather than treating a Nvidia guarantee as a red flag, are more inclined to read it as validation: it means the tenant's obligations are backed by the strongest credit in the industry, lowering SoftBank's own risk on the building[19].
The numbers show how sensitive SoftBank's stock is to this single relationship. On July 28, SoftBank shares fell more than 6% and chip supplier Kioxia dropped more than 18%, dragging Japan's Nikkei index lower[19]. Earlier in 2026, SoftBank had already fallen more than 12% on reports that OpenAI's IPO might be delayed[21]. For SoftBank, the OpenAI relationship isn't one investment among many — it's close to a single point of failure.
There's a third party in this with a much more local stake and far less say: Pike County, Ohio. Supporters see federal land, dormant since the uranium plant closed, put back to work with construction jobs and tax revenue[8][11]. Local opponents argue that 9.2 gigawatts of new gas generation and $4.2 billion in grid upgrades are being built for one private tenant, and that residents will carry the environmental and rate costs whether or not the AI boom lasts[12].
Reading the Same Numbers, Reaching Different Verdicts
Coverage of Monday's report split less over the facts than over what the facts were assumed to prove. Reuters led with the hedge, putting "WSJ reports" in its own headline and stating plainly it couldn't verify the story — a choice that's cautious by design, even if it leaves less room for analysis of what the deal would actually do[1].
Business-focused outlets moved faster to a verdict. CNBC ran a straight news piece alongside a companion analysis calling the possible deal "another strike against the AI trade," a phrase that treats the case against AI valuations as already accumulating before any contract exists[2][5]. Bloomberg's widely quoted "$750 billion in deals" headline combined the unsigned $250 billion OpenAI guarantee with a separate, already-announced roughly $500 billion partnership with a different company, SK Group — two distinct deals folded into one round number that reads as more concentrated than either is alone[3].
Right-leaning, market-skeptical outlets went further. ZeroHedge called the project "gargantuan," and Tech Times ran the headline that the guarantee "Proves Debt Markets Said No" — treating an unconfirmed term sheet as a settled verdict[7][16]. On the left, NBC News framed the story around a wider web of Nvidia-OpenAI-AMD deals creating what it called "a mirage of growth," emphasizing contagion risk over the legal mechanics of what a lease guarantee actually is[17]. In Asia, coverage stayed closer to the ledger: Japanese and Chinese-language outlets focused on what the deal means for SoftBank's balance sheet and chip suppliers like Kioxia, treating it as a portfolio-risk story rather than an American governance scandal[19].
What's Actually Left Open
Strip away the framing on every side, and two things remain true regardless of who turns out to be right. The physical project is real: federal land in Ohio, a 10-gigawatt design, billions in planned gas and transmission work, and a first phase targeted for 2028[8][11]. The financing that would pay for it is not yet real: it rests on one newspaper's sourcing, which a second wire service says it cannot confirm, and which people close to the talks say might still collapse[1].
Between those two facts sits the actual question nobody can answer yet. If the guarantee gets signed, the credit risk doesn't disappear — it moves from a wide pool of lenders onto Nvidia's single balance sheet, where it stays invisible as debt until the day it's actually called. If OpenAI's revenue grows into its costs, the guarantee never gets triggered and never costs Nvidia a dollar. If it doesn't, Nvidia would owe lease payments on a campus at a scale roughly 71 times the lease-guarantee exposure it currently discloses to the SEC[6].
That's the trade nobody can settle from where things stand today: whether a guarantee this size is Nvidia clearing the last real bottleneck in its own growth, or Nvidia quietly becoming the shock absorber for a bet the rest of the credit market wouldn't take on its own. The report itself says the talks could still fail. Until papers are signed or the deal is dropped, that's where it stays.
Summary
On Monday, July 27, 2026, The Wall Street Journal reported that Nvidia is in talks to guarantee roughly $250 billion of financing for OpenAI[1][2]. The money would let OpenAI lease a huge data-center campus in Pike County, in southern Ohio[2]. The campus is being developed by SB Energy, a unit of Japan's SoftBank[2]. A guarantee is not a loan. Nvidia would not hand over cash. It would promise lenders and the landlord that if OpenAI cannot make its payments, Nvidia will[1]. That promise matters because OpenAI does not have an investment-grade credit rating, and lenders would price the debt off Nvidia's much stronger credit instead[3][7]. Nvidia is separately said to be discussing up to $350 billion more to finance OpenAI's chip purchases[1][3].
Markets reacted fast. Nvidia shares fell about 5% on Monday, to around $197[5]. The cost of insuring Nvidia's debt against default jumped 14 basis points in one day, to 82 basis points — the biggest single-day rise since that contract started trading actively[6]. In plain terms: insuring $10 million of Nvidia debt for a year went from roughly $68,000 to roughly $82,000. The level is still modest. The speed of the move is what got attention.
The fight is over a term that carries the whole dispute: "circular financing." Critics, including investor Michael Burry, say Nvidia is funding the customers who buy Nvidia chips, so some of Nvidia's reported demand is really Nvidia's own money coming back[15]. They point to the 1990s telecom bust, when Lucent and Nortel lent to shaky customers and later wrote off billions[14]. Nvidia rejects this. It sent Wall Street analysts a memo saying it does not do vendor financing[13]. Chief executive Jensen Huang has argued its stakes are small next to what those companies raise on their own[4]. Defenders add a structural point: Lucent booked revenue and carried the customer's IOU on its books, while Nvidia's exposure here is a contingent promise on a lease, not a loan it has already made[14].
Nothing is signed. Reuters said it could not verify the report, and sources cautioned the deal could still fall apart[1]. The single most important open question is not whether the buildout is big — everyone agrees it is — but whether Nvidia's guarantee moves real risk off the market's books or simply concentrates it on one balance sheet.
The Event
On July 27, 2026, The Wall Street Journal reported that Nvidia was in talks to guarantee about $250 billion of lease and construction financing so OpenAI could occupy a planned 10-gigawatt data-center campus in Pike County, Ohio, developed by SoftBank subsidiary SB Energy[1][2]. The report said Nvidia was separately discussing financing of up to $350 billion for OpenAI's purchases of Nvidia chips, and that the full project could cost more than $500 billion including the chips[1][3]. Nvidia shares fell about 5% during Monday trading, to roughly $197, and other AI hardware stocks fell with them[5]. Nvidia's five-year credit default swap spread rose 14 basis points on the day to 82 basis points, its largest one-day increase since the contract began trading actively[6]. Reuters reported that it could not independently verify the Journal's account and that people familiar with the talks cautioned a deal might not happen[1].
Undisputed Facts
- The Wall Street Journal reported on July 27, 2026 that Nvidia was in talks to guarantee roughly $250 billion of financing tied to an OpenAI data-center lease[1][2].
- Reuters reported it could not independently verify the report, and that the talks could still fail to produce a deal[1].
- The campus in question is a planned 10-gigawatt project in Pike County, Ohio, on federal Department of Energy land at the former Portsmouth uranium enrichment site[8][9].
- OpenAI, SoftBank Group and SB Energy publicly announced a partnership to build and operate AI data centers under the Stargate program[10].
- Nvidia's five-year credit default swap spread rose 14 basis points on July 27 to 82 basis points, the largest one-day rise since the contract began actively trading[6].
- Nvidia shares fell about 5% on July 27, to roughly $197, leading a broader decline in AI hardware stocks[5].
- OpenAI does not hold an investment-grade credit rating[7].
- Nvidia has sent Wall Street analysts a memo stating that it does not engage in vendor financing, and Jensen Huang has publicly disputed the claim that its customer investments are circular[4][13].
- Powering the Ohio campus is planned to require about 9.2 GW of new natural gas generation plus roughly $4.2 billion of transmission work with AEP Ohio, with a first phase of about 800 MW expected in 2028[8][11].
The Pressure
Strip away the moralizing and blame. What structural realities persist regardless of which narrative wins?
- Nvidia must keep the buildout financed to keep selling
- Nvidia's growth depends on someone building the buildings and the power plants that hold its chips. Those projects are slow and capital-hungry, and the biggest buyer of frontier compute is not creditworthy on its own[1][7]. Lending its credit rating is the cheapest lever Nvidia has: it costs no cash today and only becomes real money if the customer fails.
- OpenAI's cash burn sets the terms
- Analyst projections cited in coverage put OpenAI's 2026 loss near $14 billion on roughly $25 billion of revenue, with no investment-grade rating[7]. That gap, not any argument about AI's future, is why a third-party guarantee is being discussed at all.
- Guarantees hide leverage in the footnotes
- A guarantee is a contingent liability. It does not show up as debt on the balance sheet until it is called. That is why credit investors watch it more closely than equity investors do, and why the reaction on July 27 was sharper in the debt market than in the stock[6].
- Physical constraints are real and slow
- The Ohio campus needs about 9.2 GW of new natural gas generation and roughly $4.2 billion of transmission work, with a first phase of about 800 MW expected in 2028[8][11]. Power plants and transmission lines take years and are approved by regulators, not by term sheets. This part of the story does not move at the speed of AI news.
- SoftBank is leveraged to the same bet twice
- SoftBank is both an equity investor in OpenAI, funded partly by a $40 billion bridge loan, and the developer of the campus OpenAI would lease[10][19]. If OpenAI stumbles, SoftBank takes the hit on both sides at once.
Material realityTwo things are true at the same time and neither depends on which narrative wins. First, the physical project is real and partly public: federal land at the former Portsmouth enrichment site in Pike County, Ohio, a 10-gigawatt design, 9.2 GW of planned gas generation, $4.2 billion of transmission work with AEP Ohio, and a first phase of roughly 800 MW targeted for 2028[8][11]. Ten gigawatts is roughly the output of ten large nuclear reactors. Second, the financing is not real yet. It rests on a single Wall Street Journal report that Reuters said it could not verify, and people familiar with the talks said it might not close[1]. Between those two facts sits the actual risk transfer. If the guarantee is signed, credit risk does not vanish — it moves from many lenders onto one company's balance sheet, where it stays off the debt line until it is called. If OpenAI's revenue grows into its costs, the guarantee is never triggered and costs Nvidia nothing. If it does not, Nvidia owes lease payments on a campus it does not use, on a scale roughly 71 times the lease-guarantee exposure it currently discloses[6].
Narrative as a weaponFour groups are actively shaping how this is read. Nvidia wants you to believe a guarantee is plumbing, not risk — hence the memo to analysts denying vendor financing and Huang's framing of its stakes as small next to what customers raise elsewhere[4][13]. Short sellers and credit bears want you to believe the loop is the story, and their most effective move is the Lucent analogy, which is vivid, historically accurate, and structurally imperfect: Lucent booked revenue and carried customer IOUs, while Nvidia's exposure here is a promise on a lease[14]. Financial media on both the left and right have converged on the same skeptical frame from different motives — systemic-risk concern on one side, distrust of financialized asset prices on the other — which means readers may mistake agreement for confirmation. And SoftBank, the party with the most to lose in two directions at once, has said the least, which is itself a choice. The most important thing being flattened in all of this is the conditional: nothing has been signed.
How Each Side Sees It
Each major actor’s view — how it frames things, its underlying incentive, and how it’s materially affected. Tap a side to read it.
Frames it asNvidia's case is that a guarantee is not a subsidy — it is how you unlock supply for a real, already-visible shortage. Its argument runs in three steps. First, the constraint on AI is power and buildings, not demand for chips; if the buildings do not get financed, Nvidia cannot sell what customers already want. Second, its credit is the cheapest tool it has: putting its name on a lease costs nothing today and only pays out if OpenAI defaults, which is far less capital-intensive than buying the data center itself[1]. Third, Huang has argued directly that Nvidia's stakes in customers are small next to the sums those firms raise elsewhere — he called the CoreWeave investment "ultimately a very small part of the amount they had to raise"[4]. Nvidia's memo to analysts makes a technical distinction its defenders think critics ignore: Lucent and Nortel extended trade credit, booked the revenue right away, and carried customer IOUs on the balance sheet. Nvidia's outbound exposure here is a contingent promise and purchase commitments, not loans already made[13][14].
WhyLock in years of guaranteed demand for its accelerators and keep the buildout on schedule, so rivals and in-house chips from cloud providers do not get an opening[1][3].
Impact on themImmediate: about 5% off the share price in one session and a record one-day rise in the cost of insuring its debt[5][6]. Longer term, a $250 billion guarantee would dwarf the lease-guarantee exposure Nvidia currently discloses — Moomoo's summary of Nvidia's fiscal first-quarter 2027 10-Q puts that disclosed cap at $3.5 billion, meaning the new commitment would be roughly 71 times larger[6].
Frames it asOpenAI's position is that owning its compute is a matter of independence, not vanity. It currently rents capacity from Microsoft, Amazon and Oracle; a leased campus of its own is the first step to controlling the thing its entire business depends on[1]. On the financing, its best argument is that a young, fast-growing company with no credit rating simply cannot borrow at reasonable rates against a 10-gigawatt project, and that this is normal for capital-heavy infrastructure — railroads, pipelines and power plants have long been built on credit support from the party with the strongest balance sheet. The counterparty risk, OpenAI's side would say, is priced by sophisticated lenders who chose to accept it.
WhySecure enough compute to keep training and serving frontier models, and reduce dependence on cloud landlords who are also competitors[1].
Impact on themWithout credit support, the project's debt would be far more expensive or unavailable. Analysts cited in coverage project OpenAI will lose about $14 billion in 2026 on roughly $25 billion of revenue, which is the core reason lenders want someone else's name on the paper[7].
Frames it asSoftBank's framing is that someone has to actually build the physical layer of AI, and that patient, concentrated capital is the only kind that can. SB Energy is developing the Ohio campus with the power generation and transmission attached — the hard, slow part nobody else was doing at this scale[8][10]. From Tokyo's vantage, a Nvidia guarantee is a validation, not a red flag: it means the tenant's obligations are underwritten by the strongest credit in the sector, which lowers SoftBank's own risk on the asset.
WhyConvert SoftBank's OpenAI bet into hard infrastructure it owns and leases, generating contracted cash flows rather than only paper gains on a private stake[10].
Impact on themSoftBank is arranging a $40 billion bridge loan tied to its OpenAI investment, with 21 additional financial institutions joining the syndication[19]. Its shares are highly sensitive to AI sentiment: SoftBank fell more than 6% on July 28 as Kioxia dropped more than 18% and dragged the Nikkei lower[19]. Earlier in 2026, SoftBank fell more than 12% on reports of a delay to an OpenAI IPO[21].
Frames it asThe skeptics' crux is not whether AI is useful. It is whether the reported demand is independent. Their argument: if the supplier guarantees the customer's rent, and the customer's main expense is buying from the supplier, then a share of that revenue is the supplier's own credit returning as sales. Michael Burry's summary was "around and around we go"[15]. Their strongest specific evidence is historical and structural. Lucent and Nortel financed loss-making telecom customers in the late 1990s; when those customers failed, Lucent had to reverse revenue and lost billions[14]. Their second point is about disclosure: a guarantee is a contingent liability. It sits in the footnotes, not in the debt total, so leverage looks lower than the real exposure. Their third is the market's own reading — the record one-day jump in Nvidia's default insurance, which they treat as bondholders repricing a risk equity holders had ignored[6].
WhyShort sellers profit if AI valuations fall; bond investors and credit analysts are paid to price downside, not upside, so their framing is structurally pessimistic[13].
Impact on themTheir view moved prices on July 27 across AI hardware names[5]. If credit spreads keep widening, Nvidia's own borrowing costs rise and the guarantee gets more expensive to extend.
Frames it asSupporters in Ohio frame the Piketon project as the reindustrialization of a county that lost its anchor employer when the federal uranium enrichment plant wound down — federal land put back to work, construction jobs, and a tax base. Opponents, organized locally, frame it differently: 9.2 gigawatts of new gas generation and $4.2 billion of transmission are being built for one private tenant, and they argue residents will bear the pollution, the water use and part of the grid cost regardless of whether the AI boom lasts[8][11][12]. Their crux is not AI finance at all — it is who pays for the infrastructure if the tenant walks away.
WhyLocal officials want jobs and revenue; opponents want to avoid ratepayer and environmental costs locked in by decades-long utility commitments[12].
Impact on themThe first phase of about 800 MW is expected in 2028[8]. A collapse in the financing would strand permitting, transmission and generation work already underway[11].
Like this article?
The Bias Ledger average rating 4
The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.
| Outlet | Vantage | Bias | How they frame it | The tell |
|---|---|---|---|---|
| Reuters | U.S./U.K. wire, center | 1 | "Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports" — attributes the claim in the headline and states plainly that Reuters could not verify it. | The hedging is the signal. By putting "WSJ reports" in the headline and flagging that the deal may fall apart, Reuters declines to treat the story as established. That is straight practice, though it also means readers get less analysis of what the structure would actually do. |
| The Japan Times | Japanese, English-language, business-establishment | 2 | "SoftBank planning massive $500 billion data center in Ohio" — earlier coverage that frames the project as a SoftBank achievement and a Japan-U.S. investment story. | The subject of the sentence is SoftBank, not the financing. Japanese coverage centers what the project means for a domestic champion's balance sheet and share price; the American argument about circular financing barely appears. |
| CNBC | U.S. center, business/investor audience | 3 | News piece: "Nvidia and OpenAI in talks for up to $250 billion backstop to fund AI infrastructure plans." A companion analysis piece: "Nvidia's potential $250B backstop for OpenAI is another strike against the AI trade." | The straight story and the verdict run side by side. "Another strike against the AI trade" reads the event through the lens of an investor already worried, and the word "another" implies an accumulating case before the deal is even confirmed. |
| Bloomberg | U.S. center, financial-professional audience | 4 | "Nvidia's $750 Billion in Deals Reignite Circular AI Fears" — combines the unsigned $250 billion OpenAI guarantee talks with a separately announced ~$500 billion SK Group partnership into one headline sum. | The $750 billion figure isn't two numbers from the same OpenAI story — it's an unsigned OpenAI guarantee added to an unrelated, already-announced SK Group deal with a different customer. Folding two distinct counterparties into one round number makes the total exposure sound more concentrated and more imminent than either deal is on its own, and it's the most quotable framing of the day, likely to be repeated without that distinction. |
| NBC News | U.S. left-of-center | 4 | Frames the story inside a wider pattern of OpenAI-Nvidia-AMD deals and the risks of AI investing becoming "insular," warning that money moving among the same firms can create "a mirage of growth." | The emphasis is systemic and public-interest — contagion, bubbles, who gets hurt. The legal mechanics of a lease guarantee, which is the point Nvidia's defense turns on, get little space, so the skeptical case reads as self-evident. |
| ZeroHedge | U.S. right-populist, market-bearish | 7 | "Nvidia Weighs $250 Billion Backstop For OpenAI's Gargantuan Ohio Data Center Campus." | "Gargantuan" does editorial work no fact requires. The site's house position is that asset prices are propped up by financial engineering, and this story is fitted to that thesis; the possibility that the guarantee is ordinary infrastructure credit support is not entertained. |
| Moomoo | Chinese-founded retail brokerage platform (Futu), Hong Kong/China; aggregates and reframes wire reports for retail traders | 7 | "NVIDIA's Credit Default Rate Surges, Is the AI Debt Bubble About to Burst?" — pairs the CDS move with the $3.5 billion lease-guarantee cap in Nvidia's fiscal Q1 2027 10-Q to argue the commitment is about 71 times the disclosed book. | The most useful specific number of the day sits under the most leading headline. "Credit default rate" is also imprecise — a CDS spread is the price of insurance, not a default rate — and "about to burst" states as imminent what the data only shows as a one-day repricing. |
References
- Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports — Reuters · International wire service, center; owned by Thomson Reuters, revenue mainly from financial data terminals
- Nvidia and OpenAI in talks for up to $250 billion backstop to fund AI infrastructure plans — CNBC · U.S. center, investor-facing; owned by Comcast/NBCUniversal
- Nvidia's $750 Billion in Deals Reignite Circular AI Fears — Bloomberg · U.S. center, financial-professional; privately held by Michael Bloomberg, revenue mainly from data terminals
- Nvidia reignites "circular" AI concerns as it weighs OpenAI financing guarantee — Axios · U.S. center, business/politics; owned by Cox Enterprises
- Nvidia's potential $250B backstop for OpenAI is another strike against the AI trade — CNBC · U.S. center, investor-facing analysis column
- NVIDIA's Credit Default Rate Surges, Is the AI Debt Bubble About to Burst? — Moomoo · Chinese-founded retail brokerage platform (Futu Holdings, Hong Kong/China); aggregates wire reports for retail traders, incentive to drive engagement
- Nvidia's $250B Guarantee for OpenAI Ohio Campus Proves Debt Markets Said No — Tech Times · U.S. tech aggregator, ad-supported, strongly editorialized headlines
- OpenAI in talks to lease 10GW data center from SB Energy in Ohio — DataCenterDynamics · U.K.-based trade publication for the data-center industry; advertiser- and event-funded, industry-friendly but technically detailed
- SoftBank eyes 10GW data center at former DOE nuclear enrichment site in Ohio — DataCenterDynamics · U.K.-based data-center trade publication
- OpenAI and SoftBank Group partner with SB Energy — OpenAI · Primary source — company announcement by a party to the deal
- Feds announce huge natural gas plant, data center project in southern Ohio — Statehouse News Bureau · Ohio public radio statehouse bureau; publicly and listener-funded, state-government focused
- Piketon, Ohio: PORTS Technology Campus 10 GW Data Center — Pike County — Stop Ohio Data Centers · Local advocacy group organized to oppose data-center development in Ohio; explicitly opposed, not neutral
- Nvidia says it isn't using 'circular financing' schemes. 2 famous short sellers disagree. — Business Insider · U.S. center-left business site; owned by Axel Springer
- Vendor Financing Loops: What 1999 Telecom Tells Us About 2026 AI — Michel Johannsen · Independent analyst blog; self-published, no disclosed institutional funding — treat as analysis, not reporting
- Nvidia Reportedly Moves to Backstop $250 Billion in OpenAI Data Center Financing, Michael Burry Says, 'Around and Around We Go' — Benzinga · U.S. retail-investor financial media; subscription and ad funded, headline-driven
- Nvidia Weighs $250 Billion Backstop For OpenAI's Gargantuan Ohio Data Center Campus — ZeroHedge · U.S. right-populist, pseudonymous, persistently bearish on asset prices and central banks
- OpenAI, Nvidia and AMD deals raise risk questions — NBC News · U.S. left-of-center broadcast news; owned by Comcast/NBCUniversal
- SoftBank planning massive $500 billion data center in Ohio — The Japan Times · Japan's largest English-language daily; business-establishment orientation, privately owned
- Kioxia Plunges 18% and SoftBank Drops Over 6% as Japan Semiconductor Sector Faces New Sell-Off — TradingKey · Asia-based retail trading analysis site; ad- and broker-referral funded
- Will OpenAI's $500 Billion Data Center Be The World's Largest? — Forbes · U.S. business magazine, center-right on markets; contributor-heavy model with uneven editing
- SoftBank shares slip over 12% on OpenAI IPO delay concerns — Nikkei Asia · Japanese business daily; owned by Nikkei Inc., which also owns the Financial Times — corporate-Japan orientation