The cloud market you thought you understood is getting a new type of competitor. xAI, the artificial intelligence company founded by Elon Musk, is increasingly discussed in industry circles as a potential “neocloud”—a model-first platform that offers hosting and inference services, essentially competing with Amazon Web Services, Microsoft Azure, and Google Cloud on AI workloads.
This is not a small shift. If model providers start offering their own infrastructure, procurement decisions get messier. The companies building the most capable AI systems would also control where and how you run them.
What Is a Neocloud and Why Does It Matter?
A neocloud is an infrastructure provider that grows out of an AI model company rather than a traditional cloud vendor. Think of it as the model maker saying, “Why rent compute from AWS when you can rent directly from us?”
xAI has been building massive GPU clusters—reportedly among the largest in the world—to train its Grok models. The logic of becoming a neocloud is straightforward: once you have that infrastructure, you can sell access to it. OpenAI already does this through its API, but xAI could go further by offering general-purpose compute and storage alongside its models.
For buyers, this creates a new category to evaluate. You are no longer choosing between hyperscalers for commodity cloud services. You are choosing between general-purpose platforms and model-native platforms that may offer better pricing or performance for specific AI tasks.
The Procurement Complexity This Creates
Traditional cloud procurement weighs factors like pricing, regional availability, compliance certifications, and ecosystem integrations. Neoclouds add new variables: model access, inference latency, and the risk of being locked into a single model provider’s infrastructure.
Consider a scenario where xAI offers Grok inference at significantly lower cost than running equivalent workloads on Azure or AWS. The savings might be attractive, but you would need to ask hard questions. What happens if xAI’s models fall behind competitors? Can you migrate workloads easily? What are the data residency and compliance guarantees?
Indian enterprises face particular considerations here. Data localisation requirements under proposed regulations could limit which platforms are viable. A neocloud without Indian data centres—or clear commitments to build them—may be a non-starter for certain workloads regardless of pricing.
How Hyperscalers Are Responding
AWS, Azure, and Google Cloud are not standing still. Each has deepened partnerships with model providers—Microsoft with OpenAI, Google with its own Gemini models, and AWS with Anthropic. They are betting that enterprises will prefer a single vendor for both infrastructure and model access.
But hyperscalers also carry their own lock-in risks. Enterprises that build heavily on Azure’s OpenAI integration may find it difficult to switch to Anthropic or Grok later without significant rework. The neocloud model, ironically, could offer more flexibility if it focuses purely on inference and lets you bring your own applications.
The market is fragmenting into tiers: hyperscalers offering broad services with model partnerships, neoclouds offering model-native infrastructure, and pure API providers offering models without infrastructure. Your sourcing strategy needs to account for all three.
Building an Evaluation Framework
If you are evaluating neoclouds alongside traditional options, start with five questions. First, what is the total cost of ownership including egress fees, which are charges for moving data out of a platform? Second, what are the contractual terms around model deprecation—if a model version is retired, what happens to your workloads?
Third, assess data residency and compliance. Does the platform meet your regulatory requirements today, and does the vendor have a credible roadmap for markets like India? Fourth, evaluate portability. Can you export models, data, and workflows to another platform without rebuilding from scratch?
Finally, consider ecosystem maturity. Hyperscalers offer thousands of integrations. A neocloud might offer better AI performance but lack connectors to your existing enterprise systems.
What This Means for You
Do not sign multi-year cloud commits without building in flexibility for this shift. The neocloud category barely existed eighteen months ago; in another eighteen months, it could represent a serious alternative for AI workloads.
Start by mapping which of your workloads are model-dependent and could benefit from model-native infrastructure. Build relationships with neocloud vendors now, even if only for pilot projects. And pressure your existing hyperscaler reps on pricing—competition from neoclouds gives you negotiating power you did not have before.
The cloud market is no longer a three-player game. Act accordingly.