Under the "iron chain" constraints, this metric is very important for tech giants.

Under the "iron chain" constraints, this metric is very important for tech giants.

```

Author: Bao Yilong

Source: Hard AI

Amid the investment boom in AI infrastructure, Remaining Performance Obligations (RPO) is becoming a key indicator for measuring future revenue, but behind this seemingly impressive number lies significant uncertainty.

On October 21, a Morgan Stanley research report pointed out that in the current AI investment frenzy, tech giants are signing long-term, large-scale contracts at unprecedented levels, creating a mutually dependent, capital-circulating “iron chain” ecosystem.

Morgan Stanley analyzes that focusing only on revenue growth and capital expenditures is no longer enough. A crucial forward-looking indicator—Remaining Performance Obligations (RPO)—is becoming key for insight into these giants’ future real income, growth quality, and potential risks.

RPO: The “Ballast” of the AI Boom

According to accounting standards, RPO (Remaining Performance Obligations) represents the portion of legally binding and irrevocable contracts that have been signed by enterprises, but for which the performance and recognition as revenue have yet to occur. It excludes optional renewals or contracts that include termination options without major penalties.

The importance of RPO lies in its offering a valuable window into future revenue trends. A growing RPO balance usually signals robust new orders and solid customer relationships. Conversely, a decline may signal early signs of slowed sales momentum or shortened contract durations.

In the AI ecosystem, several key companies have seen explosive growth in RPO balances. Data show that in the past six quarters:

Microsoft’s RPO grew by 55%.Coreweave’s RPO grew by 218%.Oracle’s RPO astonishingly grew by 411%.

Notably, Oracle disclosed at its latest investor day that the company gained about $65 billion in incremental RPO from seven contracts with four clients, pushing its total RPO past the $500 billion mark. In comparison, Google and Amazon’s RPOs are much smaller.

(RPO balances of major hyperscale cloud providers)

Not All RPOs Are Created Equal

The report points out that although RPO provides visibility on future revenue, investors must delve into its composition, as RPO quality differs significantly across companies.

First is the large disparity in contract durations. Oracle and Coreweave's massive RPOs mainly come from multi-year, long-term contracts. Meanwhile, Microsoft, Amazon, and Google’s contract periods are comparatively shorter, with revenues recognized closer to when contracts are signed.

Through the ratio “RPO/Past 12 Months’ Revenue,” we see Coreweave and Oracle's ratios are as high as 14.4x and 8.5x respectively, while Microsoft is only 1.3x, and Amazon and Google are around 0.3x.

(RPO/Past 12 Months’ Revenue: Coreweave as high as 14.4x)

High ratios mean these companies have locked in demand for many years in advance in their current RPO. This may lead to a slowdown in new contract growth even if future potential business activity remains steady.

More importantly, long-term contracts face higher renegotiation risks. If future client demand or pricing dynamics change, businesses may be forced to make concessions on pricing, terms, or service volume in order to maintain client relationships.

As Coreweave disclosed among its risk factors, if clients are no longer willing to sign long-term “take-or-pay” contracts in future, it will negatively impact their cash flow forecasts, profit margins, and business prospects.

Reassessing Valuation and Risk: The Truth Revealed by RPO

Comparing RPO to a company’s market capitalization can more clearly reveal its impact on valuation.

Data show that, including the latest disclosed contracts, RPOs account for as much as 81% and 60% of the market caps of Coreweave and Oracle, respectively. In contrast, this same ratio is only 9% for Microsoft, and even lower for Amazon and Google.

(RPO as a percentage of Coreweave and Oracle’s market value reaches 81% and 60%)

This huge gap suggests that investors have attributed most of Coreweave and Oracle’s value to these signed future revenues. For Microsoft, Amazon, and Google, however, their valuations reflect more of their vast ecosystem’s as-yet-unsigned, broader growth opportunities.

Additionally, two major uncertainties remain:

Profit and execution risk: For companies like Coreweave and Oracle, whose RPO growth is driven by emerging businesses, the costs required to fulfill these large-scale contracts (including infrastructure, personnel, and maintenance capital expenditures) are highly uncertain. This directly affects the net returns from these contract revenues.Client concentration risk: Relationships among players in the AI ecosystem are increasingly intertwined, forming an “iron chain” as client and supplier roles increasingly overlap.

(Capital flows within the AI ecosystem)

An Exception: Nvidia’s “Alternative” Model

In stark contrast to the aforementioned companies’ surging RPOs, despite Nvidia announcing many AI-related collaborations, its RPO balance has remained at a relatively low level of about $1.8 billion on average over the past year.

The report notes this reflects Nvidia's unique “optional procurement” business model, rather than long-term locked-in commitments.Before orders are formally placed, neither buyer nor seller bears contractual obligations.

This is also attested by Nvidia’s management, who have said investors should not be overly obsessed with the “10GW target” mentioned in agreements with OpenAI—the actual figures could vary greatly.

This model brings Nvidia flexibility, but also means its future revenue visibility cannot be reflected through the RPO metric.

In conclusion, for investors focused on the AI field, RPO is a powerful tool for measuring the certainty of future growth. But it’s necessary to look beyond the surface numbers and carefully scrutinize the underlying contract durations, profit margins, execution risks, and client structure.

This article is from the WeChat public account “Hard AI”. For more cutting-edge AI news, click here

Risk Warning and DisclaimerThe market involves risks, investment requires caution. This article does not constitute personal investment advice and does not take into account individual users' special investment objectives, financial status, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Invest accordingly and at your own risk. ```