Barclays downgrades Oracle's debt rating: Cash may run out by November next year, potentially becoming "junk bonds."

Barclays downgrades Oracle's debt rating, warning of debt risks.

On November 11, Barclays Bank's fixed income research department released a report stating that Oracle's capital expenditures to fulfill its massive AI contracts have far exceeded its free cash flow capacity, forcing it to rely heavily on external financing.

The report predicts that Oracle will face a severe financing gap starting from the 2027 fiscal year (beginning June 2026), and the company may run out of cash by November 2026.

Analyst Andrew Keches downgraded Oracle's debt rating to "Underweight," equivalent to a "sell" recommendation. He also warned that Oracle could ultimately fall to a BBB- rating, close to junk bond territory.

Barclays believes that the entire hyperscale vendor industry is funding the AI race by issuing massive bonds, which has begun to put pressure on the credit market. Oracle's predicament is particularly pronounced, with a debt-to-equity ratio as high as 500%, while Amazon's is 50%, Microsoft's is 30%, and Meta and Google's are even lower.

Wall Street Journal previously mentioned that on Monday, JP Morgan's strategist team also warned that the trillion-dollar capital demand from the AI boom will "squeeze" every credit market.

(Funding sources for AI/Data Centers)

Surge in Capital Expenditures, Tight Cash Flow

The direct reason for the surge in capital expenditures is the skyrocketing costs of building AI data centers.

Barclays cites industry data indicating that the construction cost of an AI data center can reach up to $50 billion to $60 billion per gigawatt, three times that of traditional data centers, with more than half of the costs allocated to purchasing computing hardware such as NVIDIA GPUs.

At the same time, capital expenditure forecasts for major companies continue to be revised upward. Since early 2025, industry forecasts for capital expenditures in the coming years have nearly doubled.

(Left: Capital expenditure forecasts for hyperscale data centers continue to rise; Right: Since the beginning of this year, the forecast for 2026 has nearly doubled.)

Data tracked by Barclays shows that in the United States alone, announced AI data center projects in the coming years will increase power demand by over 45 gigawatts, corresponding to over $2 trillion in investment.

Barclays points out that free cash flow is no longer the only source of funding.

Although most hyperscale vendors, aside from Oracle, can still generate substantial free cash flow, large-scale stock buybacks and dividend plans from companies like Google and Meta have significantly reduced the cash available for capital investments From the perspective of "net free cash flow" after deducting shareholder returns, the funding gap has become more pronounced, which also explains the recent motivation for Meta, Google, and Oracle to issue large amounts of bonds consecutively.

Debt issuance by hyperscale vendors has "flooded" the market

Barclays pointed out in its report that although the capital expenditure plans of hyperscale vendors have long been no secret, the leading companies in the industry previously relied mainly on their strong operating cash flow to support expansion, rarely engaging in large-scale financing in the bond market.

However, this model has recently been broken.

According to Barclays statistics, in the past few months, several major hyperscale vendors have issued a total of $140 billion in bonds through public and private markets, with the annual issuance expected to reach $160 billion.

(The amount of borrowing to support AI data center construction has seen a sharp increase since September)

These transactions are massive, with an average single transaction amount reaching $25 billion. Notably, even for AA-rated Meta and Google, the spreads on their newly issued bonds have significantly widened, indicating that the market is demanding a higher risk premium to absorb such a large supply.

Barclays believes this is not a one-time phenomenon. As the AI race heats up, capital expenditure forecasts and financing needs are expected to continue to grow. Although the issuance pace may slow in the short term, "longer-term, higher-level" debt issuance will become the new normal.

The report suggests that behind this trend is the cash flow tightness faced by companies like Meta, Google, and Oracle. In contrast, Microsoft and Amazon have more robust cash positions, giving them greater flexibility and patience in financing.

Oracle becomes the "weakest link," with huge financing needs

Among all hyperscale vendors, Oracle's financial situation is the most fragile.

Barclays' analysis shows that Oracle is the only company with negative free cash flow, and its debt-to-equity ratio has reached 500%, far exceeding Amazon's 50% and Microsoft's 30%.

Barclays' sensitivity analysis predicts that even if capital expenditures are not further increased, Oracle's cash will be exhausted by November 2026, indicating a significant refinancing need for the company.

If capital expenditures grow further as the market expects, Barclays' own model predicts that its fiscal year 2027 capital expenditures will be 50% higher than market consensus, leading to an even more astonishing funding gap.

Additionally, Oracle has over $100 billion in lease commitments off its balance sheet that have yet to be recognized, which will further heighten rating agencies' concerns about its leverage levels. In contrast:

• Meta, while also facing a cash gap and potentially continuing to be a bond issuer, has a substantial liquidity buffer of about $80 billion
• Google can still maintain over $70 billion in liquidity in most cases, with little financing pressure in the short term.
• Amazon and Microsoft still have positive net free cash flow under the most extreme capital expenditure scenarios, with no significant financing needs.

(The relationship between capital expenditures and net debt issuance of large-scale cloud computing companies)

Based on the above analysis, Barclays has downgraded Oracle's debt rating to "underweight." The report predicts that Oracle's credit rating may eventually be downgraded to BBB-, which is the lowest tier of investment grade.

Companies with this rating and facing ongoing cash consumption may see their spreads align with other high-risk BBB- issuers (such as some automotive and cable companies).

Barclays analyst Andrew Keches stated in the report that Oracle will face significant financing needs starting in fiscal year 2027 (beginning in June 2026), which will limit the performance of its secondary market bonds.

Meanwhile, Oracle's growth largely depends on supplier financing agreements signed with clients such as OpenAI, which increases its counterparty risk exposure.

As these partners seek to hedge risks, demand for Oracle's CDS will rise. Barclays also believes that given Oracle's partnership with OpenAI worth up to $300 billion, its CDS can also be seen as a concern regarding OpenAI-related risks.

(Oracle's 5-year CDS surges)