Signing a "divorce timetable"! Microsoft and OpenAI have a "friendly breakup", bringing the era-defining "century-long alliance" in AI to an end.

Microsoft and OpenAI have announced a restructuring of their partnership. The "union of the century" that began in 2019 has officially entered the "amicable separation" phase, as both sides retain the core framework of cooperation while embarking on new paths of independent development.

On October 28 local time, Microsoft and OpenAI announced the restructuring of their cooperation, officially ushering in the "finale" of the "union of the century" that launched the era of AI.

From Microsoft's first investment in OpenAI in 2019 to the current signing of a "divorce timetable," the "union of the century" has lasted for five years. Now, this "amicable separation" continues the core cooperation framework while defining clear paths for each party's independent development.

As Microsoft’s official blog states: "Entering the next stage of the partnership, both companies are more capable than ever of continuing to build outstanding products that meet real-world needs."

In response, Wall Street's two leading investment banks—JP Morgan and Morgan Stanley—quickly released research reports, but their interpretations of this "amicable separation" reveal subtle differences.

JP Morgan: Uncertainty Eliminated, Microsoft Welcomes a "Moment of Relief"

JP Morgan analyst Mark R Murphy pointedly noted in his report that the greatest significance of this restructuring agreement is the "reduction of key uncertainties for Microsoft stock." In his view, Microsoft’s stock stagnation in recent months was not due to fundamentals, but because "known unknowns relating to recent adjustments to the OpenAI agreement have affected investor visibility."

Murphy specifically emphasized the importance of extending intellectual property authorization to 2032. Under the previous framework, once OpenAI announced the realization of AGI (Artificial General Intelligence), Microsoft's IP authorization would immediately terminate—this was like a "Sword of Damocles" hanging over Microsoft. The new agreement not only extends the authorization period to 2032, but more importantly introduces an independent expert validation mechanism, providing an objective means of arbitration for defining AGI.

"This gives Microsoft a seven-year window of certainty: even in some AGI scenarios, Microsoft's Copilot product line and Azure OpenAI service can ensure business continuity."

JP Morgan calls Microsoft’s commitment to provide OpenAI with $250 billion worth of Azure cloud computing services "a positive surprise." Previously, news that OpenAI had reached a $300 billion compute partnership with Oracle triggered concerns about Microsoft’s potential declining status. Murphy believes this hefty commitment "should be seen as a significant rebalancing of the scales."

However, JP Morgan also noted Microsoft’s key concession: giving up its right of first refusal as OpenAI’s preferred compute provider. Murphy’s interpretation is somewhat incisive—this may stem from Microsoft’s own risk management considerations.

"Our guess is that Microsoft is using safeguards relating to customer concentration, and we would not want to see an agreement in which 75% of Microsoft’s RPO/backlog comes from a single entity."

From a financial health perspective, $250 billion may already be near Microsoft’s risk management ceiling. JP Morgan would prefer to see a longer contract to lower the annual risk exposure from a single customer.

JP Morgan characterizes the agreement as "a reasonable compromise." Microsoft gains long-term technological authorization for seven years and a massive cloud order, while OpenAI secures the flexibility to work with third parties and space to transition to a for-profit entity.

Especially noteworthy is that OpenAI may now "jointly develop non-API products with third parties," "release open weight models," and "provide API access to US government national security clients, regardless of the cloud provider."

Murphy believes the last provision echoes the Stargate project, allowing OpenAI to operate independently in sensitive areas like national security.

JP Morgan maintains its "Overweight" rating on Microsoft, with a target price of $565, based on an approximately 35x 2026 PE multiple. Murphy summarizes:

"Clarity on future structure should be viewed as a positive, helping OpenAI and Microsoft achieve their respective goals, whether jointly or independently."

Morgan Stanley: Core Framework Retained, But an Era of Competition and Cooperation Arrives

Compared with JP Morgan’s "risk elimination" view, Morgan Stanley analyst Keith Weiss’s team pays more attention to the strategic shift behind the agreement. They point out at the start of their report that the agreement "marks a turning point for the deep partnership that began in 2019," signaling "a new landscape where tech giants shift from alliance to competitive cooperation in the AGI race."

The Weiss team stresses that the key change in the new agreement, while continuing the existing cooperative framework, lies in "independence." The agreement explicitly states that Microsoft can now independently or with third parties pursue AGI development—a major breakthrough. However, if Microsoft uses OpenAI's IP to develop AGI, the relevant models will be subject to compute threshold limits, "thresholds significantly higher than the current system scale used by leading training models."

Morgan Stanley provides a more detailed interpretation of IP terms. The Weiss team notes that Microsoft's IP rights are divided into two categories:

Model and Product IP: Extended to 2032, includes post-AGI models, but comes with safeguards;

Research IP: Continues until an expert panel validates AGI or 2030, whichever comes first.

Notably, research IP does not include model architecture, model weights, inference code, fine-tuning code, nor IP related to datacenter hardware and software—Microsoft retains rights to these non-research IPs. Weiss stresses in the report:

"The extension of Microsoft’s IP rights may be an underestimated benefit. Microsoft can now utilize these IP rights across its entire business, and these rights are extended to 2032 and include post-AGI models."

Regarding the same $250 billion Azure contract, Morgan Stanley’s interpretation varies. The Weiss team believes the deal "marks a major shift in the commercial relationship between the two parties," putting Microsoft on equal footing with Oracle as one of OpenAI’s core cloud service providers."

Morgan Stanley expects the contract to significantly increase Microsoft's commercial bookings and remaining performance obligations (RPO) in the second quarter of fiscal 2026 (ending December). The analyst maintains previous forecasts for OpenAI’s contribution to Azure AI and expects the new contract will substantially boost revenue under OpenAI’s "costs."

Unlike JP Morgan’s "uncertainty elimination" view, Morgan Stanley lists several unresolved questions:

Revenue sharing mechanism: Specific details of accounting treatment and arrangements for "spreading payment over a longer term";Contract details: Duration of the $250 billion contract, intended uses (inference, training, or cross-workload), etc.;AGI timeline: This timing will determine when Microsoft’s potential "OpenAI revenue cliff" might arrive;Backlog size: The true size of OpenAI’s pro forma backlog.

The research report especially highlights that the time of AGI realization becomes a crucial variable affecting the relationship. Because the revenue sharing arrangement will continue until AGI is validated by an independent expert panel, "this might occur before or after the previously forecast 2030 partnership end date."

Morgan Stanley maintains its "Overweight" rating and $625 target price (higher than JP Morgan’s $565), listing Microsoft as its top pick in the software sector.

Risk Warning and DisclaimerThe market involves risks, and investment should be made cautiously. This article does not constitute personal investment advice and does not consider the unique investment goals, financial situation, or needs of any individual user. Users should consider whether any opinions, viewpoints, or conclusions in this article fit their particular circumstances. Any investments made accordingly are at the user's own risk.