Microsoft’s AI Growth Is Becoming Increasingly Dependent on Big Tech
The AI industry is entering an interesting phase. While companies across manufacturing, transportation, finance and other sectors are beginning to experiment with artificial intelligence, a significant share of the spending on AI infrastructure and models is still coming from the technology industry itself.
A recent development involving Microsoft and Meta Platforms highlights this trend.
According to people familiar with the matter, Meta is spending hundreds of millions of dollars each year to access AI models through Microsoft’s Azure cloud platform. Meta reportedly consumes trillions of tokens every week through the platform, making it one of Microsoft's largest AI customers.
Both companies declined to comment on the figures.
Why Meta is spending so heavily on external AI
Meta is already one of the largest companies developing its own AI models and infrastructure. So why would it need to spend hundreds of millions of dollars accessing models from another company?
The answer is relatively straightforward: developing AI is not only about building your own model.
Meta's developers use several different AI models and platforms depending on their availability, capabilities and cost. External models can also be useful for evaluating Meta's own systems.
For example, developers at Meta have reportedly used OpenAI technology through Microsoft's Foundry platform to assess the performance of Meta's own AI models.
This creates an interesting situation where an AI company can simultaneously be:
Developing its own models
Buying access to competitors' models
Using cloud infrastructure from another technology company
Building infrastructure to eventually reduce its dependence on external providers
Meta's CTO Andrew Bosworth has also discussed the company's use of multiple leading AI models during the development process.
Microsoft Foundry and the AI marketplace
Microsoft's strategy extends beyond its own Copilot products and AI models.
Through Azure AI Foundry, Microsoft provides customers with access to models from different AI providers. The platform reportedly had around 100,000 customers as of July.
On paper, this gives Microsoft an important position in the AI ecosystem. Instead of customers having to work with individual model providers separately, they can access multiple models through Microsoft's cloud infrastructure.
However, there is an important detail behind those numbers.
Although Microsoft increasingly highlights customers from traditional industries such as manufacturing and transportation, people familiar with the business say many of its largest AI customers are still technology companies.
ByteDance, the company behind TikTok, has reportedly been one of the largest spenders on Foundry. Other significant customers reportedly include Adobe, Perplexity and AI startup Sierra.
That raises a bigger question about the current economics of AI.
Is AI spending becoming too concentrated within the technology sector?
The AI boom has produced enormous amounts of investment and spending across the technology industry.
But there is a potential weakness in this model.
If major technology companies are simultaneously developing AI infrastructure, buying AI services from one another and selling AI-related products to each other, some of the reported growth can become concentrated within the same ecosystem.
This does not mean the revenue is necessarily artificial or that the underlying demand does not exist.
These companies are genuinely spending significant amounts of money on computing power, models, data centers and software.
The question is whether this spending eventually expands beyond the technology sector.
For AI to become a broader economic transformation, companies in industries such as manufacturing, healthcare, logistics, financial services and transportation will need to move from experimentation to meaningful, large-scale adoption.
Microsoft has another concentration problem
Microsoft's dependence on technology companies is not limited to Foundry.
OpenAI remains one of Microsoft's most important AI relationships. According to the information in the report, OpenAI accounted for roughly 70% of Microsoft's overall AI revenue in its most recent fiscal year.
Microsoft's AI business therefore has exposure to a relatively small number of extremely large customers.
That creates both an opportunity and a risk.
Large customers can generate substantial revenue quickly, but relying heavily on a few companies can make future growth more sensitive to their spending decisions.
Meta is also trying to reduce that dependence
There is another interesting part of the story.
Meta is reportedly building its own business for providing access to different AI models through an API service.
If successful, such a platform could eventually compete with Microsoft's Foundry.
This would give Meta another potential source of revenue while also providing greater control over its AI ecosystem.
It could also reduce Meta's reliance on external platforms for accessing AI models.
This reflects a broader pattern emerging across the technology industry: companies are simultaneously cooperating and competing with each other.
A company can be a customer of another AI provider today while developing a competing product tomorrow.
The bigger picture
The Microsoft–Meta relationship illustrates something important about the current AI market.
The AI economy is not simply a story of individual companies building independent models. It is increasingly an interconnected ecosystem involving cloud providers, model developers, chip companies, software platforms and large enterprise customers.
Microsoft provides infrastructure and access to models.
Meta develops its own models while also consuming external ones.
OpenAI provides models while depending heavily on Microsoft's infrastructure.
Other technology companies purchase these services and, in some cases, develop competing platforms of their own.
The result is a rapidly evolving network of partnerships and competition.
The next important stage may be determining whether this spending eventually spreads much further into the broader economy.
If businesses outside the technology sector begin deploying AI at scale—and generating measurable improvements in productivity, revenue or cost efficiency—that could provide stronger evidence that the current AI investment cycle is translating into a wider economic transformation.
For now, however, the numbers suggest that Big Tech remains one of the biggest engines driving AI demand.
And that may be one of the most important things to watch as the AI industry moves into its next phase.
#ArtificialIntelligence #AI #Microsoft #Meta #Azure #GenerativeAI #CloudComputing #Technology #AIInfrastructure #TechIndustry #DigitalTransformation #BusinessTechnology #FutureOfAI