Oracle Corporation

KBRA released a report on the credit implications of AI infrastructure expansion among the AI-7 companies—Meta, Amazon, Alphabet, Microsoft, NVIDIA, Broadcom, and Oracle—finding that their gross disclosed contractual commitments and contingent support have surged to approximately $3.2 trillion from $575 billion at the end of 2024. The report highlights that these off-balance sheet commitments, including long-term leases, purchase and construction commitments, and cloud capacity agreements, are reshaping the companies' financial risk profiles and reducing future flexibility, despite generally strong balance sheets. Six of the AI-7 maintain traditional lease-adjusted leverage below 1.5x, but the rapid growth in commitments is a concern. KBRA emphasizes that commitments are not economically uniform and should be analyzed separately based on timing, utilization, demand, cancellability, and counterparty performance.

OpenAI's head of data centers, Chris Malone, departed the company last week, becoming the fourth senior executive to leave ahead of the company's planned 2027 listing, the Wall Street Journal reported. Malone joined OpenAI in March 2025, shortly after the company unveiled Stargate, a joint venture with Oracle and SoftBank to build AI data-center capacity at scale. His exit follows the departures of Chief Revenue Officer Denise Dresser, Chief Operating Officer Brad Lightcap, and Fidji Simo, who served as second-in-command to CEO Sam Altman. Oracle, a core partner in the Stargate venture, saw its shares edge lower Tuesday on the news. OpenAI has raised its projected spending on computing power to $750 billion through 2030, up from an earlier estimate of roughly $600 billion, and this month signed a 10-gigawatt data-center lease in Ohio with SoftBank's SB Energy, with Nvidia agreeing to provide a guarantee of up to $105 billion to support that lease.

Paramount Skydance is facing pushback from a collection of US attorneys general in its proposed acquisition of rival Warner Bros. Discovery. Media analyst Evan Shapiro says the deal will remain in limbo unless Paramount backs out, and that Larry Ellison is personally backing $40 billion dollars of this $111 billion dollar deal. Shapiro notes Oracle stock has dropped 60% since bidding began, and if Oracle's debt is rated junk, Ellison's personal stock gets locked up, putting the Ellisons in financial hot water. He adds that Netflix, Apple, or a spun-out NBC could swoop in if Paramount walks away, which he thinks is increasingly likely given a $7 billion breakup fee.

Dan Ives says Nvidia's roughly 15% price increase on AI server systems is not a demand warning but another sign that AI infrastructure remains severely supply constrained. He told Bloomberg Television that demand and supply for chips is upwards of 15 to one, with the current memory shortage forcing Nvidia and others to pass rising component costs through the supply chain. Ives expects the memory super cycle to keep the industry from reaching meaningful supply-demand equilibrium until the middle or latter part of 2028. Microsoft, Alphabet and Oracle are reportedly among customers facing higher data-center system costs as memory prices rise, yet Ives sees little evidence that those costs are causing customers to retreat. He described the AI buildout as still being in the third inning, with enterprise adoption accelerating rather than slowing, and expects hyperscalers to continue raising capital and increasing capital expenditures.

Nvidia plans to raise prices on servers containing its artificial intelligence chips by more than 15% in many cases, with the increases set to take effect on systems shipped early next year, according to Bloomberg. Vera Rubin and Grace Blackwell chip-based systems are among those subject to the increases, with the exact percentage varying by chip generation and how much memory a given configuration includes. Contract server manufacturers that supply major data center operators including Microsoft, Alphabet's Google, and Oracle have recently passed word of the upcoming price adjustments to their customers. The driver behind the hikes is the rising cost of memory chips, which are a core component of Nvidia's GPUs and server systems, as explosive growth in AI infrastructure spending has outrun the ability of the three dominant DRAM producers to supply the market. Nvidia posts a gross margin of 75%, and its chips command prices in the tens of thousands of dollars apiece, reflecting persistent shortfalls in supply from contract manufacturer Taiwan Semiconductor Manufacturing Co. relative to what buyers need.

Market strategists are raising pointed concerns about the traditional safe-haven status of U.S. Treasuries as the national debt crosses $40 trillion. Eduardo Repetto, chief investment officer at Avantis Investors, said in a CNBC interview that when you get to 40 trillion, that's a big number, so maybe it's not as safe as we thought. David Rosenberg, founder and president of Rosenberg Research, argued the real culprit behind rising yields since February is a surge in long-dated issuance by hyperscalers, with three of the four major hyperscalers now running free cash flow negative and forced to tap capital markets. Repetto also noted that benchmark bond indexes are debt-weighted, so the flood of Treasury issuance has shifted allocations without investors realizing it, with government securities now around 50% of the Bloomberg index versus 20% twenty-five years ago. Rosenberg said Treasury Secretary Scott Bessent's recent $4 billion bond buyback was merely a trial balloon, and that the government has drawn a line in the sand that it's not going to allow the ten-year to get to 5%.

Major technology companies are facing a pivotal shift from the AI boom to proving who can actually generate profits. After the latest earnings reports, investors are beginning to question who has enough money to keep funding massive investments. Reuters estimates that Big Tech will spend more than 700 billion dollars on capital expenditure in 2026. Alphabet has raised its 2026 capital expenditure forecast to between 195 billion and 205 billion dollars, and posted negative free cash flow of 5.9 billion dollars in the second quarter. Meanwhile, Meta's free cash flow fell 91 percent to just 784 million dollars, with full-year capital expenditure expected at 130 billion to 145 billion dollars. Amazon, Alphabet, Meta and Oracle have issued bonds totaling about 194 billion dollars this year, up 79 percent from a year earlier. Tencent accelerated its latest quarterly capital expenditure to 52.8 billion yuan, up from 31.9 billion yuan in the previous quarter. Investors should therefore change how they view AI stocks, focusing on companies with pricing power that can generate revenue from AI faster than their costs and convert capital expenditure back into real free cash flow.

Big Tech companies are deploying aggressive public relations campaigns to counter growing local opposition to AI data centers across the US. At least 75 data center projects worth about $130 billion combined were blocked or delayed by local opposition during the first three months of this year, according to research group Data Center Watch, approaching the total value of projects stalled by local resistance in all of 2025. Oracle has unleashed one of the most aggressive pro-data center PR efforts to rally support for its massive 2.45 gigawatt Project Jupiter facility in New Mexico, a centerpiece of its $300 billion computing deal with OpenAI. Meta Platforms has spent millions on TV commercials, Microsoft said it would no longer seek tax breaks from communities where it builds data centers, and OpenAI is meeting with residents and pledging to conserve water and pay its own way for power. The opposition has been particularly prevalent in political battleground states like Michigan and Georgia, where 17 projects have been delayed or blocked, and governors in New York, Texas, and Pennsylvania have issued executive orders pausing or restricting new data center permits.

SoftBank Group is preparing to issue approximately 1 trillion yen, or $6.26 billion, in seven-year corporate bonds targeting Japanese retail investors, which would be the largest such sale ever by a Japanese company. The offering comes as Nvidia's 5-year credit default swap spread climbed to 80.77 basis points as of August 19, surpassing its late-July peak and rising roughly 90% year-to-date. Long-term government yields in the U.S., Germany, and Japan have reached their highest levels in decades, with investors citing hyperscaler bond supply as a contributing factor. AI infrastructure debt now accounts for as much as 30% of net new issuance in some investment-grade indexes this year, according to private credit market commentary cited by Reuters. Japan's broader retail-targeted corporate bond market is on pace to hit a record 2.8 trillion yen this year, per Nikkei, but SoftBank's offering would dwarf anything attempted before. Oracle offers the starkest credit warning in the sector, with its 5-year CDS spread surpassing 200 basis points in July 2026, approximately four times the investment-grade corporate index and a level not seen since the 2008 financial crisis.

A Reuters report says Microsoft, Alphabet, Amazon, Meta and Oracle are on track to spend more on combined investment than the free cash flow they generate by 2027. Operating cash flow at the five companies is expected to rise by about 340 billion dollars, while annual capital expenditure will increase by roughly 534 billion dollars. That means every one dollar of additional cash flow comes with about 1.57 dollars of extra investment. A clear example is Oracle, where cloud infrastructure revenue grew 77 percent in fiscal 2026, but the company posted negative free cash flow of 23.7 billion dollars and had to raise money through both debt and equity to support data center construction. Meanwhile, the IEA estimates that global data center electricity use will rise from about 485 terawatt hours in 2025 to 950 terawatt hours in 2030, and New York has become the first US state to pause environmental permits for new data center projects using 50 megawatts or more for one year.

Morgan Stanley says the artificial-intelligence boom is entering a more capital-intensive phase, raising a new question for investors in Microsoft, Alphabet, Amazon, Meta Platforms and the companies supplying their infrastructure: not whether AI spending will continue, but how increasingly enormous investment plans will be financed. The bank expects combined capital expenditures at the four hyperscalers to jump 57% in 2027 from 2026, as companies race to add data centers, computing capacity and power infrastructure. Morgan Stanley believes the economics can justify that spending, estimating AI investments can generate returns on invested capital above 25%, but the problem is timing. AI infrastructure requires cash upfront, while revenue and free cash flow arrive later, and Morgan Stanley's 2027 free-cash-flow estimates for the four hyperscalers have continued falling, creating what the bank sees as a widening financing gap. Credit markets are already reacting, with hyperscaler debt spreads widening roughly 35 basis points for higher-quality borrowers and about 50 basis points for lower-rated companies at their summer peak, though spreads have narrowed sharply over the past two weeks. Cash-rich hyperscalers and suppliers including Nvidia and Broadcom have considerably more financing flexibility, while companies further down the credit spectrum, including Oracle, data-center developers, REITs and former bitcoin miners, could become more sensitive to rising borrowing costs. The financing wave is also expanding beyond buildings, as Morgan Stanley expects private capital and asset-backed structures to increasingly fund servers, chips and energy infrastructure, pointing to Nvidia's compute-infrastructure financing platform and a $35 billion Broadcom-backed chip financing transaction.

Steve Eisman, the investor known for his 'Big Short' bet against subprime mortgages, says the AI boom has an 'Achilles' heel' in its enormous dependence on OpenAI and Anthropic, but he is not ready to short the sector yet because investors lack evidence that a breakdown has begun. Speaking on his Weekly Wrap podcast, Eisman recalled that before shorting subprime, his team bought access to Moody's securitization database and watched delinquency data deteriorate month after month, confirming his thesis, but he said there is no such data set with respect to AI because both labs are private. He sees a real bear case in that there don't seem to be any moats around large language models, as users switch between models constantly and Chinese open-weight rivals are far cheaper, which could eventually trigger a price war. Research estimates cited by Eisman suggest OpenAI and Anthropic account for roughly 70% of AI revenue across Microsoft, Amazon and Alphabet, and 25% to 35% of their cloud revenue, while roughly half of the approximately $600 billion backlog at Oracle comes from OpenAI. Eisman called that dependency huge and quite scary, warning that if a price war undermines the labs' economics, hyperscalers may pull back on capex and the entire AI chain goes into reverse. He said when OpenAI and Anthropic go public, we will have some real data, and prediction-market traders expect Anthropic to reach public markets first, with Polymarket giving Anthropic a 90% chance of beating OpenAI to an IPO and a 70% chance of completing its IPO before Nov. 1. So far, Eisman sees little evidence that the breaking point has arrived, as hyperscalers are still spending heavily and Nvidia is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilize more than $500 billion for AI infrastructure, while Anthropic is projecting $190 billion to $200 billion of 2028 revenue, up from a $47 billion annualized run rate in May.

Nine tech giants now carry roughly $3 trillion in off-balance-sheet AI obligations, nearly double a July estimate of $1.65 trillion and triple their reported debt. A Wall Street Journal analysis found about $1.9 trillion in purchase commitments and $1.2 trillion in unstarted leases across Microsoft, Amazon, Alphabet, Meta, Oracle, Nvidia, Broadcom, Advanced Micro Devices, and SpaceX. Alphabet alone disclosed $811 billion in purchase commitments as of June 30, up 152% from $322 billion three months earlier. Meta's Hyperion data center project in Louisiana carries about $27 billion in debt that never touches its balance sheet because Blue Owl Capital funds own the majority of the joint venture. Deferred depreciation could exceed $520 billion over three years, potentially pushing Oracle's depreciation as a share of revenue from 7% toward 28% and Meta's from 9% toward 19% by fiscal 2028.

Oracle is reportedly preparing another round of layoffs as the software giant tries to contain payroll costs while pouring tens of billions of dollars into AI infrastructure. Some teams could face double-digit percentage reductions, with managers asked to identify employees for cuts ahead of Oracle's fiscal second quarter, which begins Sept. 1. The company's workforce has already fallen to roughly 141,000 employees from 162,000 a year earlier. The cuts come despite booming demand, as fiscal 2026 revenue rose 17% to a record $67.4 billion and cloud revenue surged 39% to $34 billion. But infrastructure spending has risen even faster, with Oracle spending roughly $42.7 billion on capital expenditures during fiscal 2026 and outlining plans to raise $45 billion to $50 billion through debt and equity to support additional cloud infrastructure.

Oracle shares dropped nearly 4% on Friday after Transwestern Pipeline, a subsidiary of Energy Transfer, said the Green Chile natural gas project in New Mexico would be delayed by six months. The pipeline is crucial to powering Oracle's massive Project Jupiter data center complex, which plans to use Bloom Energy fuel cells to supply up to 2.5 gigawatts of electricity. Transwestern revised the in-service date to February 1, 2027, from the original August 15, 2026, citing repeated denials by the state over routing on public land. An Oracle spokesman told Bloomberg that Project Jupiter remains on schedule and the company continues to work closely with partners.

Oracle disclosed plans to raise roughly $40 billion in fiscal 2027 as its shares fell about 1.7% Friday morning. Half could come from Oracle's already announced $20 billion at-the-market equity program, while management does not expect to issue more debt during calendar 2026. Remaining performance obligations rocketed 363% year over year to $638 billion, with another $85 billion piling on in just one quarter. Operating cash flow jumped 54% to $32 billion in fiscal 2026, yet free cash flow still crashed to negative $23.7 billion as spending exploded. Oracle closed at $153.14 on August 14 versus a GF Value estimate of $190.69, putting the shares 19.69% below GF Value.

Oracle's stock has fallen nearly 50% from its September 2025 peak amid a credit downgrade, rising bond insurance costs, and doubts about its massive AI investment. S&P Global Ratings cut Oracle's long-term issuer credit rating to BBB- from BBB in July, leaving it one notch above junk, while Moody's holds a negative outlook. Oracle's total liabilities ballooned from $147.4 billion to $218.7 billion in twelve months, and free cash flow for fiscal 2026 was negative $23.7 billion. On the fourth quarter earnings call, Chief Executive Clay Magouyrk addressed rising component costs directly, stating: 'So I don't like it when costs go up. Our customers don't like it when costs go up. And honestly, I don't think our suppliers do.' Chief Financial Officer Hilary Maxson said Oracle expects to raise around $40 billion in debt and equity in fiscal year 2027, including a previously announced $20 billion equity issuance.

Michael Burry has warned that $879 billion in hyperscaler purchase commitments circulate through Nvidia, inflating reported revenue without multiplying actual customer demand. The investor who shorted subprime mortgages before the 2008 crash shared a Bloomberg diagram tracing roughly $46 billion in direct equity stakes and $879 billion in multi-year purchase commitments among Microsoft, Oracle, Amazon, Google, Meta Platforms, OpenAI, Anthropic, xAI, CoreWeave, Nvidia, and Advanced Micro Devices. The Bank for International Settlements estimates the five largest hyperscalers carry about $1.65 trillion in off-balance-sheet debt through special purpose vehicles, exceeding the $1.35 trillion they report directly, and Nvidia's five-year credit default swap spread has roughly doubled over the past two months. Burry's post suggests the same dollar can show up as revenue at more than one stop on the chain, with hyperscalers funding AI labs that then spend on compute from those same hyperscalers. Nvidia sits at the center of the web with a $5.4 trillion valuation, while Anthropic and OpenAI are worth roughly $1.8 trillion combined despite neither being profitable.

Michael Burry added to his short position in Nebius Group on August 12, the same day the AI cloud company's stock surged 34% after reporting second-quarter revenue of $582.3 million, up 454% year-over-year and ahead of analyst estimates of $572.75 million compiled by LSEG. Burry, who first disclosed his NBIS short on August 6 at approximately $212 per share, called Nebius "what the top of a boom looks like" in a Substack post, and also raised short stakes in Micron and Oracle. Nebius, the Nvidia-powered AI cloud provider spun out of Yandex and headquartered in Amsterdam, reported AI cloud revenue jumped 514% to $575 million, closed four landmark deals averaging more than $1 billion each in total contract value, and raised its 2026 contracted power target to 5 gigawatts from 4 gigawatts. Burry's bear case centers on depreciation accounting and off-balance-sheet commitments, arguing AI infrastructure companies are stretching the useful life of chips and overstating earnings. Nebius ended June with $8 billion in cash and generated $2.2 billion in operating cash flow, though capital expenditures of roughly $5.7 billion exceeded analyst estimates of $4.7 billion.