Large companies are limiting AI capacity for startups

The world's largest cloud service providers – giants such as Microsoft, Amazon, and CoreWeave – are reserving graphics processor capacity mainly for their internal projects and major partners such as Anthropic and OpenAI. This was reported by Zamin.uz.

As a result of this approach, small and medium-sized AI startups are forced to wait in multi-month queues to gain access to computing power. As industry specialists emphasize, this situation negatively affects the pace of market development.

Due to a shortage of computing resources, the rental prices of these services have increased by more than twenty-five percent in the last six months. According to experts' forecasts, this tense situation is expected to last at least until the end of 2026.

Today, many newly established technology enterprises point to the shortage of graphics processors as the main obstacle on the path of their development. A three-tier hierarchy for distributing GPUs has been introduced in the Microsoft system, in which the largest and most prestigious clients have priority.

For example, a client wishing to rent new-generation Blackwell chips is required to order at least one thousand units and sign a long-term contract worth tens of millions of dollars. Such conditions are almost impossible for small companies and are squeezing them out of the market.

Due to the worsening situation on the market, some startups are switching to building their own infrastructure in order not to depend on external providers. In particular, the Collide company announced that it would invest five hundred thousand dollars in creating its own computing cluster in order to operate independently and rid itself of the resource shortage.

Such steps may reduce the dependence of small companies on large cloud platforms in the future.