PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteThe AGI clause threatened to make OpenAI and Microsoft’s partnership most contentious just as the technology behind it became most valuable. It tied major rights—including revenue sharing, cloud access and intellectual-property arrangements—to a milestone whose full contractual definition was not public. The companies’ April 27, 2026 amendment shifted key commercial terms toward fixed dates and broader cloud flexibility. That reduced the chance that one disputed AGI determination would force an abrupt split, but it did not end the partnership.
What the AGI clause did—and what “secret” gets wrong
The clause was not wholly secret: OpenAI and Microsoft publicly described its broad effects, and regulators summarized parts of the relationship. What was not public was the complete contract language, including the operational definition of AGI and all the procedures for deciding whether the contractual threshold had been met.
In the October 28, 2025 agreement, OpenAI could declare that it had achieved AGI, but an independent expert panel had to verify the declaration. That distinction mattered: a company announcement alone was not the contractual determination. According to OpenAI’s summary of the agreement, expert verification affected the end of revenue sharing and certain research-IP rights. Microsoft’s rights to OpenAI models and products, by contrast, were extended through 2032, including for post-AGI models.
The public summary also distinguished research IP from other IP. Certain research-IP rights could end at expert verification or in 2030, whichever came first. Microsoft retained rights to non-research IP identified in the announcement, including model architecture, weights, inference code and fine-tuning code, as well as data-center hardware and software. The Microsoft filing described the company’s exclusive IP and Azure API rights as continuing until AGI, alongside the extended model and product rights.
#1 Best Overall
So AGI was not a simple switch that would automatically cut Microsoft off from every OpenAI model or product. Different rights had different terms, and the public materials do not disclose the full legal mechanics.
Why a scientific milestone became a commercial trigger
“AGI” has no single universally accepted test. It can refer to broad human-level performance, economically valuable automation, autonomous completion of complex work or a threshold specified in a private contract. Strong benchmark scores would not necessarily establish that a system met the contractual test; an economically useful system might also raise questions even if it did not resemble a general human intelligence.
That ambiguity had unusually high stakes because the parties were simultaneously investors, infrastructure partners, licensors, distributors and commercial collaborators. An AGI determination could affect revenue sharing, research access, cloud exclusivity and the parties’ ability to pursue other arrangements. The UK Competition and Markets Authority’s partnership summary also describes the relationship’s IP, AGI-related provisions and mutual revenue sharing.
Rank #2
The contractual process was meant to constrain unilateral action, but independent verification did not eliminate uncertainty. Public announcements did not explain who selected or paid the experts, what evidence they could review, how confidential model information would be handled, whether the outcome required unanimity, or how a disputed or inconclusive finding would be resolved. Those details should not be assumed.
Why each company had reason to resist a badly timed determination
OpenAI’s incentives
OpenAI had reasons to seek more room to secure compute, distribution and partners beyond Microsoft, while avoiding a commercial transition governed by a hard-to-predict AGI finding. This is an inference from the companies’ announced arrangements, not evidence of a private plan to conceal or accelerate AGI.
The October 2025 restructuring paired Microsoft’s investment—then valued at approximately $135 billion, or roughly 27% of OpenAI Group PBC on an as-converted diluted basis—with OpenAI’s commitment to purchase an additional $250 billion of Azure services. Those figures, reported by OpenAI in its partnership announcement, show why greater flexibility and continued Microsoft integration could both matter.
Microsoft’s incentives
Microsoft had reasons to protect the value of its investment and preserve dependable access to OpenAI technology for Azure and its products. Independent verification limited the risk that OpenAI could declare AGI on its own and immediately trigger contractual consequences. The October 2025 filing also set out Microsoft’s exclusive IP and Azure API rights until AGI, while the public agreement summary described model and product rights through 2032.
Microsoft later characterized its rights as including royalty-free access to OpenAI IP through 2032, according to its FY2026 third-quarter earnings materials. These were meaningful protections, but they did not make every category of IP or every commercial right identical.
Why cloud flexibility became a pressure point
OpenAI’s interest in additional infrastructure and distribution was not theoretical. Its November 2025 announcement described an AWS infrastructure partnership and a $38 billion compute commitment. In February 2026, OpenAI and Amazon announced a broader strategic partnership involving AWS infrastructure, Amazon Bedrock distribution, OpenAI Frontier, Trainium capacity, customized models and a potential $50 billion Amazon investment. The announcements are described by OpenAI’s AWS partnership notice and its Amazon partnership announcement.
Cloud arrangements involve more than where a model runs. Training and inference capacity, API hosting, first-party products, third-party distribution and managed agent services can be governed differently. In February 2026, OpenAI and Microsoft said Azure remained the exclusive cloud provider for stateless OpenAI API calls, including calls arising from third-party collaborations, under the terms then in force. They also said the contractual AGI definition and determination process were unchanged. See their joint statement.
That chronology matters: Amazon’s partnership did not itself show that Azure had been displaced. It did, however, make the limits of a highly exclusive arrangement more commercially consequential.
What changed in the April 2026 amendment
On April 27, 2026, OpenAI announced a revised agreement that moved several headline commercial terms away from AGI-dependent transitions and toward fixed dates, a cap and broader cloud access. The announcement did not explicitly say that the AGI clause was deleted; the safer reading is that key economic and exclusivity mechanics were redesigned. The announced terms are in OpenAI’s amendment summary.
Best Value
| Issue | October 2025 terms as publicly described | April 2026 amendment as publicly described |
|---|---|---|
| AGI process | OpenAI could declare AGI, subject to independent expert verification; key rights and revenue sharing were linked to verification. | The announcement shifted major commercial mechanics to fixed dates and a cap; it did not state that the AGI provision itself was deleted. |
| Microsoft IP license | Exclusive IP and Azure API rights were described as lasting until AGI; model and product rights extended through 2032. | Microsoft’s IP license became non-exclusive and continues through 2032. |
| Revenue sharing | Revenue sharing continued until expert verification, according to the public summary. | Microsoft stopped paying revenue share to OpenAI. OpenAI’s payments to Microsoft continue through 2030, independent of technological progress and subject to a cap. |
| Cloud access | Azure held an exclusive role for stateless APIs under the February 2026 statement, while third-party partnerships were contemplated. | OpenAI can serve products across cloud providers. Microsoft remains the primary cloud partner, and products ship first on Azure unless Microsoft cannot or chooses not to support the required capability. |
The practical trade was less exclusivity in exchange for greater predictability. OpenAI gained room to serve products on other clouds; Microsoft retained a long-dated, now non-exclusive license and a primary-cloud position. The parties also continued to describe work together on infrastructure, silicon, cybersecurity and products in the April announcement.
The change became operational quickly: on April 28, OpenAI said its models, Codex and managed agents were coming to AWS in limited preview. That announcement establishes a preview launch, not general availability or feature parity across every OpenAI service. See OpenAI’s AWS availability notice.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the amendment fixed—and what remains unsettled
The amendment reduced the importance of a single AGI decision to the headline commercial relationship. Fixed dates and a payment cap are easier to plan around than an uncertain technical trigger. But it did not establish that all AGI-related terms vanished, and the full amended contract was not included in the public announcement.
- The partnership did not end. Microsoft remains a major shareholder and primary cloud partner, while OpenAI can work with other cloud providers.
- Microsoft did not lose access altogether. Its license is non-exclusive, not terminated, and is announced to continue through 2032.
- Azure and AWS are not interchangeable in every arrangement. The relevant product, hosting path, contractual counterparty and availability status can differ.
- Some obligations persist. OpenAI’s payments to Microsoft continue through 2030 under the announced capped framework.
- The companies still have overlapping interests. Microsoft is investor, supplier, distributor and licensee; OpenAI seeks infrastructure and market flexibility.
For organizations buying OpenAI capabilities, the amendment is a reason to compare delivery paths, not to assume one is cheaper or universally preferable. A buyer should check model and feature availability, data controls, identity and networking, procurement terms, support responsibilities and whether a service is in preview. The contract changes principally concern distribution, cloud flexibility and strategic dependence; they do not establish a price advantage for Azure, AWS or direct access.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why the clause could have broken the old model
A partnership can benefit both sides while a technology is being built, then become harder to sustain when that technology changes the value of the bargain. The AGI clause put that transition into a contract: a contested determination could have affected access, economics and cloud rights at once. Independent verification reduced unilateral control but could not make an undisclosed definition or process certain to both parties.
The April 2026 amendment appears to have lowered that fault line by substituting more time-based and capped commercial terms for some AGI-linked mechanics. It preserved a close relationship while making it less exclusive. The central risk was not that AGI automatically meant separation; it was that the milestone could force the companies to renegotiate who controlled and benefited from the technology at precisely the moment their incentives diverged.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
