Microsoft and OpenAI have rewritten the terms of one of the technology industry’s most consequential AI partnerships, trading strict exclusivity for a broader, more flexible operating model. The amended agreement keeps Microsoft Azure at the center of OpenAI’s infrastructure strategy while allowing OpenAI to make its products available across other cloud providers. The result is not a breakup, but a recalibration: Microsoft keeps deep access and economic upside, while OpenAI gains more room to scale, sell, and compete in a market that is moving faster than any single cloud can comfortably absorb.
The revised partnership reflects a larger truth about the current AI boom: frontier AI has become an infrastructure problem as much as a software problem. Training and serving advanced models now requires enormous compute capacity, specialized chips, global data center buildouts, and reliable enterprise distribution. No single agreement from 2019 or 2023 could fully anticipate the scale of demand that followed the rise of ChatGPT, Microsoft Copilot, and enterprise generative AI deployments.
Microsoft and OpenAI’s relationship began as a strategic bet. Microsoft supplied capital, cloud infrastructure, and go-to-market power; OpenAI supplied breakthrough research and models that helped redefine the consumer and enterprise AI landscape. That combination gave Microsoft a first-mover advantage in integrating generative AI into Windows, Microsoft 365, GitHub, Azure, security products, and developer platforms.
The new arrangement preserves much of that foundation but removes some of the rigidity. Microsoft remains OpenAI’s primary cloud partner, and OpenAI products are still expected to ship first on Azure unless Microsoft cannot or chooses not to support the necessary capabilities. At the same time, OpenAI can now serve products across other cloud providers, creating a more open distribution model for customers that already operate on AWS, Google Cloud, Oracle Cloud, or hybrid infrastructure.
Key changes include:
Microsoft’s original advantage came from being OpenAI’s essential infrastructure and commercialization partner. That role helped Azure become nearly synonymous with enterprise AI, especially for organizations that wanted access to OpenAI models through Microsoft’s compliance, identity, and security layers. The amended deal keeps that advantage but acknowledges that OpenAI’s ambitions now exceed any one cloud pipeline.
For OpenAI, the shift is a sign of maturity. A company serving consumers, developers, governments, and global enterprises cannot afford to make every customer’s deployment strategy dependent on one cloud vendor. That may have worked during the early platform-building phase, but it becomes constraining when AI becomes a mainstream business utility.
The new structure creates three layers of partnership:
The non-exclusive license may sound like a downgrade, but Microsoft has spent the past several years embedding AI throughout its own portfolio. Microsoft 365 Copilot, GitHub Copilot, Azure AI Foundry, Windows AI features, and security copilots all benefit from model access, enterprise relationships, and integrated workflows. Microsoft’s defensibility increasingly comes not only from exclusive access to models, but from how deeply those models are wrapped into productivity, identity, data governance, and developer tools.
There is also a risk-management element. By reducing obligations to OpenAI and keeping long-term access through 2032, Microsoft creates more room to diversify its AI model supply. The company has already signaled interest in a broader model ecosystem, including internal models and third-party alternatives. That does not diminish OpenAI’s importance, but it reduces the danger of strategic overdependence.
The new model lets OpenAI pursue customers where they already are. A bank standardized on Google Cloud, a retailer deeply invested in AWS, or a government agency with a multi-cloud procurement mandate may be more willing to buy OpenAI products if the deployment path fits existing architecture. This matters because AI adoption is increasingly shaped by compliance, latency, data residency, procurement policy, and operational familiarity.
The agreement also helps OpenAI reduce strategic friction. It can pursue cloud capacity, data center partnerships, chip initiatives, and regional deployments without every move appearing to test the boundaries of Microsoft’s rights. That flexibility is especially important as AI companies race to secure power, GPUs, custom accelerators, and long-term hosting commitments.
This changes the competitive conversation from exclusive access to execution. Microsoft will need to prove that Azure delivers the best blend of price, performance, reliability, compliance, and integration for OpenAI workloads. AWS can pitch scale and enterprise breadth. Google Cloud can emphasize AI-native infrastructure, TPUs, and model ecosystem depth. Oracle can compete on high-performance infrastructure and strategic capacity deals.
For customers, the shift could be beneficial. Multi-cloud availability may reduce lock-in concerns and create more negotiating leverage. It may also improve latency and resilience if OpenAI services can be deployed closer to users or within preferred regional cloud environments.
Under a more flexible OpenAI distribution model, those customers can evaluate AI deployment based on architecture rather than vendor politics. That could accelerate adoption in industries where cloud standardization is already locked in. It may also make OpenAI easier to approve through procurement because buyers can align purchases with existing vendor management and compliance programs.
For WindowsForum readers, the enterprise Windows angle is important. Microsoft still has a powerful advantage through Windows, Entra ID, Microsoft 365, Defender, Purview, GitHub, and Azure. Even if OpenAI products appear elsewhere, Microsoft can offer a deeply integrated productivity and security layer that rivals cannot easily replicate.
But choice comes with complexity. A model served through Azure may not behave identically in operational terms to the same or similar model served elsewhere. Rate limits, logging options, latency, regional availability, security features, and integration hooks can differ. Developers will need to pay attention to the platform around the model, not just the model name.
This also raises questions about portability. Teams that build tightly around one provider’s SDK, identity framework, vector database, monitoring system, or agent orchestration tool may still face lock-in even when the underlying model is available elsewhere. Multi-cloud access does not automatically equal multi-cloud portability.
That matters because AGI is not a clean contractual trigger. The term is contested, difficult to measure, and vulnerable to strategic interpretation. If a major business relationship depends on whether one party declares or verifies AGI, legal and commercial uncertainty can overshadow product planning.
By emphasizing access through 2032 and revenue-sharing through 2030, the companies are choosing a more predictable framework. Microsoft can plan around model access. OpenAI can plan around payments and infrastructure flexibility. Customers can plan around platform availability rather than speculative debates about whether a model has crossed a philosophical threshold.
This has several implications:
Data centers have become the new strategic frontier. AI workloads require dense power, specialized cooling, high-speed networking, and enormous capital expenditure. Cloud providers are racing to secure energy contracts, build regional capacity, and deploy accelerators fast enough to meet demand. Microsoft has the balance sheet and operational experience to remain a critical partner for OpenAI, even in a non-exclusive world.
Chip development is equally important. Nvidia GPUs remain central to the AI boom, but every major player wants more control over cost and supply. Microsoft’s internal chip efforts, OpenAI’s hardware ambitions, and the broader shift toward custom accelerators all point to a future where model performance depends as much on silicon strategy as software architecture.
Security opportunities include:
The amended structure could help Microsoft argue that OpenAI is not locked into Azure. It could help OpenAI argue that it has more independence and broader market access. But regulators may still ask whether Microsoft’s equity position, IP rights, product integrations, and enterprise distribution give it disproportionate influence over the AI market.
Competitors will likely use the change in two ways. They will claim the end of exclusivity proves the market needs openness, while also rushing to capture newly available OpenAI demand. That makes the agreement both a concession to competition and a catalyst for more intense rivalry.
If OpenAI can use more cloud providers, it may improve service resilience and reach. That could matter during peak demand, major product launches, or regional expansion. More infrastructure options may also support specialized experiences, such as multimodal tools, real-time voice, video generation, or AI agents that require heavy inference capacity.
For Microsoft users, Copilot’s future remains closely tied to the broader Microsoft ecosystem. Windows, Edge, Office apps, Teams, Xbox services, and security products can still benefit from OpenAI technology under Microsoft’s long-term license. The bigger question is whether Microsoft increasingly blends OpenAI models with its own models and other third-party systems behind the scenes.
OpenAI now has more room to operate like a global AI platform company. That means more commercial opportunity, but also more responsibility. It must maintain product consistency, safety standards, enterprise trust, and infrastructure reliability across a broader set of partners.
What to watch next:
Source: Social Samosa Microsoft and OpenAI revise partnership terms to scale AI
Overview
The revised partnership reflects a larger truth about the current AI boom: frontier AI has become an infrastructure problem as much as a software problem. Training and serving advanced models now requires enormous compute capacity, specialized chips, global data center buildouts, and reliable enterprise distribution. No single agreement from 2019 or 2023 could fully anticipate the scale of demand that followed the rise of ChatGPT, Microsoft Copilot, and enterprise generative AI deployments.Microsoft and OpenAI’s relationship began as a strategic bet. Microsoft supplied capital, cloud infrastructure, and go-to-market power; OpenAI supplied breakthrough research and models that helped redefine the consumer and enterprise AI landscape. That combination gave Microsoft a first-mover advantage in integrating generative AI into Windows, Microsoft 365, GitHub, Azure, security products, and developer platforms.
The new arrangement preserves much of that foundation but removes some of the rigidity. Microsoft remains OpenAI’s primary cloud partner, and OpenAI products are still expected to ship first on Azure unless Microsoft cannot or chooses not to support the necessary capabilities. At the same time, OpenAI can now serve products across other cloud providers, creating a more open distribution model for customers that already operate on AWS, Google Cloud, Oracle Cloud, or hybrid infrastructure.
Why the change matters now
This is a practical response to scale, not merely a legal tweak. OpenAI needs more compute, more routes to customers, and more commercial flexibility. Microsoft needs predictability, a clearer return profile, and the ability to keep building its own AI stack without being seen as dependent on one partner.Key changes include:
- Microsoft keeps access to OpenAI model and product intellectual property through 2032
- That license becomes non-exclusive
- OpenAI can offer products across multiple cloud platforms
- OpenAI products remain Azure-first under defined conditions
- Microsoft no longer pays revenue share to OpenAI
- OpenAI continues revenue-sharing payments to Microsoft through 2030, subject to a cap
- Microsoft continues to participate in OpenAI’s growth as a major shareholder
The Partnership Is Looser, But Not Weaker
The biggest misread would be to frame this as Microsoft losing OpenAI. The agreement still gives Microsoft a privileged position in OpenAI’s ecosystem, including continued model access, Azure-first treatment, and equity exposure. What changes is the exclusivity structure that made the relationship feel like a single-lane highway.Microsoft’s original advantage came from being OpenAI’s essential infrastructure and commercialization partner. That role helped Azure become nearly synonymous with enterprise AI, especially for organizations that wanted access to OpenAI models through Microsoft’s compliance, identity, and security layers. The amended deal keeps that advantage but acknowledges that OpenAI’s ambitions now exceed any one cloud pipeline.
For OpenAI, the shift is a sign of maturity. A company serving consumers, developers, governments, and global enterprises cannot afford to make every customer’s deployment strategy dependent on one cloud vendor. That may have worked during the early platform-building phase, but it becomes constraining when AI becomes a mainstream business utility.
From exclusivity to priority
The phrase primary cloud partner matters. It means Microsoft remains first in line for major OpenAI launches, but not necessarily the only channel through which OpenAI reaches customers. This is a more nuanced position than full exclusivity and more defensible in a market where regulators, enterprises, and rivals are watching AI platform concentration closely.The new structure creates three layers of partnership:
- Azure-first product launches when Microsoft can support the needed capabilities
- Non-exclusive distribution that lets OpenAI reach customers on other clouds
- Long-term IP access that lets Microsoft continue integrating OpenAI technology into its own products
Microsoft’s Strategic Trade-Off
For Microsoft, the amended agreement looks like a calculated exchange: give up exclusivity, gain predictability. The company no longer has to pay OpenAI a revenue share, while OpenAI continues payments to Microsoft through 2030 at the same percentage rate, subject to a total cap. This reshapes the economics in Microsoft’s favor in some respects, even as it opens the door to more competition.The non-exclusive license may sound like a downgrade, but Microsoft has spent the past several years embedding AI throughout its own portfolio. Microsoft 365 Copilot, GitHub Copilot, Azure AI Foundry, Windows AI features, and security copilots all benefit from model access, enterprise relationships, and integrated workflows. Microsoft’s defensibility increasingly comes not only from exclusive access to models, but from how deeply those models are wrapped into productivity, identity, data governance, and developer tools.
There is also a risk-management element. By reducing obligations to OpenAI and keeping long-term access through 2032, Microsoft creates more room to diversify its AI model supply. The company has already signaled interest in a broader model ecosystem, including internal models and third-party alternatives. That does not diminish OpenAI’s importance, but it reduces the danger of strategic overdependence.
What Microsoft gains
The amended terms give Microsoft several concrete advantages:- A clearer revenue-sharing endpoint through 2030
- No ongoing obligation to share its own OpenAI-related revenue with OpenAI
- Continued model and product IP access through 2032
- A major shareholder position in OpenAI’s future growth
- Freedom to deepen relationships with other AI model providers
- A stronger argument that Azure is preferred, not locked-in by contract
OpenAI Gets the Scale It Needs
OpenAI’s side of the bargain is straightforward: the company needs more distribution and more compute flexibility. Frontier AI is capital-intensive, and inference demand can grow unpredictably when products become popular. If ChatGPT, enterprise APIs, agents, multimodal tools, and custom business applications all scale simultaneously, a single-cloud arrangement becomes a bottleneck.The new model lets OpenAI pursue customers where they already are. A bank standardized on Google Cloud, a retailer deeply invested in AWS, or a government agency with a multi-cloud procurement mandate may be more willing to buy OpenAI products if the deployment path fits existing architecture. This matters because AI adoption is increasingly shaped by compliance, latency, data residency, procurement policy, and operational familiarity.
The agreement also helps OpenAI reduce strategic friction. It can pursue cloud capacity, data center partnerships, chip initiatives, and regional deployments without every move appearing to test the boundaries of Microsoft’s rights. That flexibility is especially important as AI companies race to secure power, GPUs, custom accelerators, and long-term hosting commitments.
A broader commercial runway
The revised deal opens several paths for OpenAI:- Selling into enterprises that prefer non-Azure clouds
- Reducing deployment friction for regulated industries
- Expanding regional availability where Microsoft capacity may be limited
- Negotiating with infrastructure partners from a stronger position
- Supporting AI products that may require specialized hosting or hardware
- Building a more independent commercial identity ahead of future financing options
Cloud Competition Enters a New Phase
The most immediate market impact is on the cloud industry. If OpenAI can offer products across multiple providers, AWS, Google Cloud, Oracle, and specialized AI infrastructure firms all gain potential openings. Azure keeps the inside track, but rivals can now compete for OpenAI workloads and customers in a more direct way.This changes the competitive conversation from exclusive access to execution. Microsoft will need to prove that Azure delivers the best blend of price, performance, reliability, compliance, and integration for OpenAI workloads. AWS can pitch scale and enterprise breadth. Google Cloud can emphasize AI-native infrastructure, TPUs, and model ecosystem depth. Oracle can compete on high-performance infrastructure and strategic capacity deals.
For customers, the shift could be beneficial. Multi-cloud availability may reduce lock-in concerns and create more negotiating leverage. It may also improve latency and resilience if OpenAI services can be deployed closer to users or within preferred regional cloud environments.
The cloud vendors’ opening
The revised agreement could reshape cloud AI competition in several ways:- AWS gains a clearer route to OpenAI-related enterprise demand
- Google Cloud can compete for customers that want OpenAI alongside Gemini
- Oracle and specialist providers can pursue infrastructure-heavy workloads
- Azure must defend its lead through integration, not exclusivity alone
- Enterprises gain more leverage in pricing and deployment negotiations
Enterprise Customers May Benefit Most
Enterprises are likely to be among the biggest beneficiaries of the revised partnership. Many large organizations already run multi-cloud environments for reasons that have little to do with AI. They may use Azure for Microsoft 365 and identity, AWS for application hosting, Google Cloud for analytics, and private infrastructure for regulated workloads.Under a more flexible OpenAI distribution model, those customers can evaluate AI deployment based on architecture rather than vendor politics. That could accelerate adoption in industries where cloud standardization is already locked in. It may also make OpenAI easier to approve through procurement because buyers can align purchases with existing vendor management and compliance programs.
For WindowsForum readers, the enterprise Windows angle is important. Microsoft still has a powerful advantage through Windows, Entra ID, Microsoft 365, Defender, Purview, GitHub, and Azure. Even if OpenAI products appear elsewhere, Microsoft can offer a deeply integrated productivity and security layer that rivals cannot easily replicate.
How IT leaders should respond
Enterprise technology teams should treat the amended agreement as a reason to revisit AI architecture plans, not as a reason to pause them.- Inventory current AI dependencies across Microsoft, OpenAI, and third-party platforms.
- Map data residency requirements against possible Azure and non-Azure deployment options.
- Review procurement contracts for model access, revenue commitments, and portability clauses.
- Benchmark latency and cost across available regions and cloud providers.
- Strengthen governance controls before expanding AI usage across business units.
- Identity integration remains easier for many Microsoft-centered organizations on Azure
- Data governance may vary depending on where OpenAI products are hosted
- Procurement leverage may improve as more cloud channels become available
- Security reviews should account for model routing and third-party infrastructure
- Cost forecasting may become more complex in multi-cloud AI deployments
Developers Get More Choice, But More Complexity
Developers may welcome the idea of OpenAI products becoming more available across clouds. Application teams often want access to the best model for a task without having to rework hosting, networking, authentication, or compliance assumptions. More deployment paths can reduce friction and make OpenAI tools easier to integrate into existing stacks.But choice comes with complexity. A model served through Azure may not behave identically in operational terms to the same or similar model served elsewhere. Rate limits, logging options, latency, regional availability, security features, and integration hooks can differ. Developers will need to pay attention to the platform around the model, not just the model name.
This also raises questions about portability. Teams that build tightly around one provider’s SDK, identity framework, vector database, monitoring system, or agent orchestration tool may still face lock-in even when the underlying model is available elsewhere. Multi-cloud access does not automatically equal multi-cloud portability.
The practical developer impact
Developers and platform engineers should watch for differences in:- API behavior and versioning
- Latency across regions
- Available model variants
- Data retention and logging defaults
- SDK support and authentication methods
- Integration with observability and security tooling
- Pricing for inference, fine-tuning, and agent workflows
The AGI Clause Becomes Less Central
One of the more important strategic consequences is the reduced role of technology milestones in the business relationship. Earlier versions of the Microsoft-OpenAI arrangement drew significant attention because of how artificial general intelligence could affect commercial rights and obligations. The revised economics appear more tied to dates, caps, and defined access periods than to uncertain declarations of technical achievement.That matters because AGI is not a clean contractual trigger. The term is contested, difficult to measure, and vulnerable to strategic interpretation. If a major business relationship depends on whether one party declares or verifies AGI, legal and commercial uncertainty can overshadow product planning.
By emphasizing access through 2032 and revenue-sharing through 2030, the companies are choosing a more predictable framework. Microsoft can plan around model access. OpenAI can plan around payments and infrastructure flexibility. Customers can plan around platform availability rather than speculative debates about whether a model has crossed a philosophical threshold.
Why predictability wins
The AI industry often talks about breakthroughs, but enterprises buy roadmaps. They need to know whether a product will be supported, whether a model will remain available, and whether a vendor relationship will survive the next research milestone. The revised agreement gives the market more calendar-based certainty.This has several implications:
- AGI debates become less disruptive to commercial planning
- Microsoft gains confidence in long-term model access
- OpenAI gains clearer payment and distribution boundaries
- Customers get a more stable basis for procurement decisions
- Regulators may find the structure easier to evaluate
Data Centers, Chips, and Security Stay Central
The companies emphasized that their collaboration will continue across data center capacity, next-generation chips, and AI-powered cybersecurity. That is a reminder that the partnership is not only about model licensing. The real battle is over the physical and operational infrastructure required to make AI reliable at global scale.Data centers have become the new strategic frontier. AI workloads require dense power, specialized cooling, high-speed networking, and enormous capital expenditure. Cloud providers are racing to secure energy contracts, build regional capacity, and deploy accelerators fast enough to meet demand. Microsoft has the balance sheet and operational experience to remain a critical partner for OpenAI, even in a non-exclusive world.
Chip development is equally important. Nvidia GPUs remain central to the AI boom, but every major player wants more control over cost and supply. Microsoft’s internal chip efforts, OpenAI’s hardware ambitions, and the broader shift toward custom accelerators all point to a future where model performance depends as much on silicon strategy as software architecture.
Cybersecurity as a shared battlefield
AI security is one area where Microsoft and OpenAI still have obvious alignment. Microsoft operates one of the world’s largest enterprise security businesses, while OpenAI models can assist with threat analysis, code review, incident response, and security automation. Together, they can turn AI into a defensive layer across endpoints, identities, cloud workloads, and developer pipelines.Security opportunities include:
- AI-assisted threat detection
- Automated incident summarization
- Secure code analysis
- Identity risk scoring
- Phishing and fraud detection
- Security operations copilots
- Policy-aware enterprise agents
Regulatory and Market Scrutiny Will Intensify
A less exclusive Microsoft-OpenAI agreement may ease some competition concerns, but it will not end scrutiny. Regulators in the United States, United Kingdom, European Union, and other markets are watching how AI power concentrates across cloud providers, model developers, chip suppliers, and application platforms. The partnership remains large enough to invite attention.The amended structure could help Microsoft argue that OpenAI is not locked into Azure. It could help OpenAI argue that it has more independence and broader market access. But regulators may still ask whether Microsoft’s equity position, IP rights, product integrations, and enterprise distribution give it disproportionate influence over the AI market.
Competitors will likely use the change in two ways. They will claim the end of exclusivity proves the market needs openness, while also rushing to capture newly available OpenAI demand. That makes the agreement both a concession to competition and a catalyst for more intense rivalry.
Antitrust questions ahead
Regulatory questions may focus on:- Whether Azure-first treatment disadvantages rival clouds
- How Microsoft’s shareholder position affects OpenAI strategy
- Whether enterprise bundles make AI competition less open
- How model access terms affect downstream developers
- Whether cloud capacity deals create indirect exclusivity
- How customer data flows across Microsoft and OpenAI systems
Consumer Impact: Subtle, But Real
Consumers may not notice immediate changes in ChatGPT, Copilot, or other AI products. The agreement is mostly about infrastructure, licensing, and commercial terms. Still, consumer-facing effects could emerge over time through reliability, regional availability, product speed, and integration choices.If OpenAI can use more cloud providers, it may improve service resilience and reach. That could matter during peak demand, major product launches, or regional expansion. More infrastructure options may also support specialized experiences, such as multimodal tools, real-time voice, video generation, or AI agents that require heavy inference capacity.
For Microsoft users, Copilot’s future remains closely tied to the broader Microsoft ecosystem. Windows, Edge, Office apps, Teams, Xbox services, and security products can still benefit from OpenAI technology under Microsoft’s long-term license. The bigger question is whether Microsoft increasingly blends OpenAI models with its own models and other third-party systems behind the scenes.
What users may eventually see
Consumer-facing changes could include:- Faster rollout of OpenAI services in more regions
- Improved reliability during demand spikes
- More model variety across Microsoft and OpenAI products
- Different capabilities between ChatGPT and Copilot experiences
- More competition among AI assistants and productivity tools
Strengths and Opportunities
The amended Microsoft-OpenAI agreement gives both companies a more realistic framework for the next phase of AI growth. It recognizes that AI scale requires flexibility, while preserving enough structure to keep the partnership strategically meaningful.The upside
- OpenAI gains broader distribution without fully severing its Azure-first relationship
- Microsoft keeps long-term model access and remains deeply embedded in OpenAI’s growth
- Enterprises gain more cloud flexibility for AI deployment and procurement
- Cloud competition intensifies, which could improve pricing and infrastructure quality
- Developers may get more deployment options across existing environments
- Regulators may view non-exclusivity as a healthier market signal
- AI infrastructure investment could accelerate across data centers, chips, and security platforms
Risks and Concerns
The deal also introduces new uncertainties. Non-exclusivity solves one problem while creating others, particularly around consistency, governance, competitive positioning, and operational complexity.The downside
- Azure may lose some exclusive pull for customers that only wanted OpenAI access
- OpenAI deployments could become fragmented across clouds with different controls
- Enterprises may face more complex compliance reviews in multi-cloud AI architectures
- Microsoft must prove its AI advantage through integration rather than exclusivity
- OpenAI may become more expensive to support operationally as distribution broadens
- Regulators may continue investigating influence and market concentration
- Customers may struggle to compare model performance and pricing across platforms
Looking Ahead
The next phase of the Microsoft-OpenAI relationship will be judged less by the announcement and more by execution. If Azure remains the best-performing and best-integrated home for OpenAI workloads, Microsoft can retain most of the strategic upside without exclusivity. If rival clouds quickly secure major OpenAI deployments, the cloud AI market could become significantly more competitive.OpenAI now has more room to operate like a global AI platform company. That means more commercial opportunity, but also more responsibility. It must maintain product consistency, safety standards, enterprise trust, and infrastructure reliability across a broader set of partners.
What to watch next:
- Whether OpenAI announces deeper cloud deals with AWS, Google Cloud, Oracle, or others
- How Microsoft prices and packages OpenAI-powered Azure services
- Whether Copilot products begin using a more diverse mix of models
- How regulators respond to the revised non-exclusive structure
- Whether enterprise buyers shift AI workloads away from Azure or simply gain negotiating leverage
Source: Social Samosa Microsoft and OpenAI revise partnership terms to scale AI
