S&P Global Ratings lowered Oracle's long-term issuer credit rating to BBB- from BBB on July 9, 2026, placing the enterprise software company one notch above speculative grade and tying the decision directly to the scale of its artificial intelligence infrastructure spending. The action also cut Oracle's short-term issuer credit rating and commercial paper rating to A-3 from A-2, with a stable outlook attached to both.

The downgrade reframes a company that marketers have known for two decades as a database and enterprise software vendor. According to S&P Global Ratings, Oracle's cloud infrastructure business accounted for 27% of revenues in fiscal 2026 and is projected to reach almost 60% of revenues by fiscal 2028. That shift, the agency wrote, moves Oracle away from the recurring, sticky enterprise revenue that supported its higher rating and toward a business model requiring heavy upfront capital with returns realized only over the length of multi-year contracts.

Why the cut happened now

S&P Global Ratings first assigned Oracle a negative outlook in July 2025, citing the pace of the AI infrastructure buildout and its potential financial impact. In the July 2026 report, the agency stated plainly that it had underestimated the situation. According to S&P Global Ratings, "we now recognize that that we underestimated the scale of the investments required to expand the AI business and its impact on our overall view of Oracle's creditworthiness."

The mechanics of the decision rest on cash flow. Oracle guided its capital expenditure to reach $90 billion to $95 billion for fiscal 2027, which ends in May. That figure sits far above the agency's prior forecast of $60 billion. According to S&P Global Ratings, most of the increase is attributable to rising component costs and new contract wins. The agency now forecasts Oracle's fiscal 2027 free operating cash flow deficit will widen to nearly $42 billion, weaker than its prior forecast of negative $24 billion. The $42 billion figure is inclusive of working-capital benefits from customer prepayments and other timing impacts, and it aligns with Oracle's own guidance of a net cash outlay for capital expenditure of around $70 billion.

Leverage tells a parallel story. S&P Global Ratings forecasts that Oracle's adjusted leverage will reach the mid-4x area in fiscal 2027. That represents an improvement over its earlier forecast of 4.8x, yet it remains above the 4x threshold that triggered the downgrade under the BBB rating. The agency credited a $5 billion mandatory convertible preferred stock issuance in February 2026 and a planned $20 billion equity issuance for the partial improvement.

The debt behind the ratios

S&P Global Ratings adjusts its debt calculation to include commitments that do not appear as conventional borrowings. According to the report, the agency added $260 billion of additional lease commitments, expected to commence between fiscal 2027 and 2029, along with unconditional purchase obligations that totaled $13 billion as of May 31, 2026. Those purchase obligations relate primarily to data center power arrangements. The lease and power figures, rather than reported balance-sheet debt alone, drive the agency's view of how stretched Oracle's finances have become.

OpenAI at the center

The single sharpest point in the report concerns customer concentration. According to S&P Global Ratings, OpenAI makes up roughly half of the $638 billion in remaining performance obligations that Oracle carries. Remaining performance obligation, or RPO, represents contracted revenue a company expects to recognize over time. A backlog of that size would normally signal strength. The agency's concern is what sits underneath it.

According to S&P Global Ratings, OpenAI's ability to meet its contractual obligations and raise external financing depends on continued AI tailwinds and its models remaining market leaders. The agency described a specific failure path: if OpenAI were unable to pay Oracle, Oracle could be left with massive data center leases that it might be unable to exit or would have to re-lease to new tenants on less favorable terms. To gauge that exposure, S&P Global Ratings said it is tracking OpenAI's financial commitments to data center operators and chip makers, treating those commitments as a proxy for OpenAI's future prospects and market share among enterprise and consumer users.

The relationship between the two companies has been building for more than a year. Oracle and OpenAI expanded their Stargate arrangement in July 2025 with plans to develop up to 4.5 gigawatts of additional capacity, part of a $500 billion initiative to build AI infrastructure across the United States. The flagship Abilene, Texas campus runs on Oracle Cloud Infrastructure. The economic logic that pushed Oracle deep into AI compute is the same logic that now concentrates its risk in a single counterparty.

A demand signal that cuts both ways

S&P Global Ratings did not dispute that demand is currently strong. According to the report, Oracle's RPO growth speaks to a robust AI compute demand environment even after recent contract terms began requiring customer prepayments. The agency framed the prepayment shift as a point in Oracle's favor, noting that better aligning cash outlays with customer prepayments lends credibility to a financial policy it now views as more conservative than a year ago.

The caution lies in duration. According to S&P Global Ratings, near-term demand could reverse if leading frontier model developers cannot raise external financing or stop subsidizing their customers. Enterprise customers could also reduce AI spending if their returns on investment underwhelm. The agency wrote that these risks do not affect industry economics today but could pressure re-leasing terms when contracts come up for renewal.

Competition reshapes the compute market

A notable element of the report addresses new entrants to the compute-leasing market. According to S&P Global Ratings, SpaceX's decision to lease its compute capacity to Anthropic and Alphabet, along with the potential for Meta to do the same, points to growing competition within the industry. The agency argued that in an industry downturn, Oracle would perform worse than other hyperscalers because it relies more heavily on external customers than on internal workloads, and because its competitors generally hold greater financial flexibility to outspend Oracle and weather a downturn.

The SpaceX arrangement referenced in the report has itself been a moving target. The compute-sharing deal with Anthropic was later clarified as a 180-day lease with a 90-day mutual cancellation provision rather than the multi-year commitment earlier reporting had described. The detail matters for the broader picture the agency paints: the terms on which compute is leased and re-leased are unsettled across the industry, and unsettled terms are precisely the pressure point S&P Global Ratings identified for Oracle's future contract renewals.

What the stable outlook rests on

Despite the cut, the agency attached a stable outlook rather than a negative one. According to S&P Global Ratings, the stable outlook reflects an expectation that Oracle will demonstrate consistent improvements toward profitability as capacity comes online and the business scales, even with stretched leverage and cash flow over the next two years. The agency expects Oracle to finance its cash-flow deficits through a mix of debt and equity in fiscal 2027 and beyond.

S&P Global Ratings also expressed a view on Oracle's intentions. According to the report, the agency believes Oracle is firmly committed to keeping an investment-grade rating, citing the $5 billion mandatory convertible preferred stock issuance in February 2026 and the planned $20 billion equity issuance in calendar-year 2026 as evidence of a more conservative financial policy. The agency expects Oracle to raise additional capital over the next two to three years, and it expects the company to issue equity beyond calendar 2026 if credit metrics were to weaken and pressure the investment-grade rating.

The path forward carries explicit conditions. According to S&P Global Ratings, it could lower the rating again if it expects Oracle to sustain leverage exceeding 4.5x, if it comes to believe Oracle is not on a path to generate positive free operating cash flow by fiscal 2029, or if it takes a negative view of Oracle's AI strategy or the overall AI industry. An upgrade, which the agency called unlikely over the next two years, would require Oracle to achieve its growth objectives while consistently generating positive cash flow and bringing adjusted leverage toward the mid-3x area on a sustained basis. The agency's base case assumes the free operating cash flow deficit will peak in fiscal 2027 and turn positive by 2029, while noting that the unit economics of the AI infrastructure business remain opaque.

Why this matters for the marketing community

The downgrade lands inside a spending pattern that has defined the advertising and technology sectors through 2026. Alphabet raised $85 billion in an equity offering backed by Berkshire Hathaway to fund AI compute, guiding capital expenditure to a range of $180 billion to $190 billion for the year, roughly six times its 2022 level. Google's own capital expenditure guidance was detailed by Sundar Pichai as a supply-constrained response to demand, with the company scaling spending from $30 billion to approximately $180 billion. Microsoft's search advertising recovery arrived alongside $80.1 billion in property and equipment additions over nine months, with roughly two thirds of one quarter's spending directed at short-lived GPU and CPU assets.

The infrastructure that serves generative AI advertising products runs on the same capital that S&P Global Ratings has now flagged as a credit risk at one of its providers. OpenAI's advertising ambitions supply the clearest thread. The company's projection of $100 billion in advertising revenue by 2030 targets the search advertising business that funds much of the wider ecosystem, and OpenAI is simultaneously the counterparty whose financing capacity the agency singled out as central to Oracle's rating. A company selling advertising inventory at scale is also the largest single obligation inside a $638 billion backlog now under ratings scrutiny.

For marketers, the practical read is about the durability of the platforms they buy from and build on. Ad tech partners increasingly plug directly into these systems, as when Kargo joined ChatGPT as an advertising technology partneralongside Adobe, Criteo, Pacvue, and StackAdapt. Cloud sovereignty and provider concentration have already surfaced as procurement questions, visible when AWS launched its European Sovereign Cloud to address data location concerns. The financial health of the companies supplying AI compute now sits in the same category of question that once applied only to a platform's product roadmap. When a ratings agency states that the unit economics of AI infrastructure remain opaque, the opacity extends to every advertising product built on top of it.

Timeline

  • July 2025: S&P Global Ratings assigns Oracle a negative outlook, citing the pace of its AI infrastructure buildout and potential financial impact.
  • July 2025: Oracle and OpenAI expand their Stargate arrangement with plans to develop up to 4.5 gigawatts of additional capacity.
  • February 2026: Oracle issues $5 billion in mandatory convertible preferred stock.
  • May 31, 2026: Oracle's unconditional purchase obligations total $13 billion, primarily related to data center power arrangements.
  • July 9, 2026: S&P Global Ratings lowers Oracle's long-term issuer credit rating to BBB- from BBB and its short-term rating to A-3 from A-2, with a stable outlook.
  • Fiscal 2027 (ending May): Oracle guides capital expenditure to $90 billion to $95 billion; S&P forecasts a free operating cash flow deficit of nearly $42 billion.
  • Calendar 2026: Oracle plans a $20 billion equity issuance.

Summary

Who: S&P Global Ratings and Oracle Corporation, with OpenAI identified as the central credit risk and SpaceX, Anthropic, Alphabet, and Meta cited as reshaping the compute-leasing market.

What: S&P Global Ratings lowered Oracle's long-term issuer credit rating to BBB- from BBB and its short-term and commercial paper rating to A-3 from A-2, placing Oracle one notch above speculative grade with a stable outlook. The agency tied the action to rising AI infrastructure spending, a forecast fiscal 2027 free operating cash flow deficit of nearly $42 billion, adjusted leverage in the mid-4x area, and OpenAI's roughly half-share of a $638 billion remaining performance obligation backlog.

When: July 9, 2026, following a negative outlook first assigned in July 2025.

Where: The rating action was published from San Francisco by S&P Global Ratings and applies to Oracle globally, with the AI infrastructure at issue centered on United States data center campuses including the Stargate site in Abilene, Texas.

Why: According to S&P Global Ratings, Oracle's growing AI infrastructure business dilutes its stronger legacy software profile through high capital expenditure, an uncertain path to profitability, a rapidly evolving competitive landscape, and high customer concentration, with the largest single obligation dependent on OpenAI's ability to raise financing and remain a market leader.