Firms Fear Shortages of Computing Power, Memory and Storage for AI
Businesses expect that within two years AI-generated workloads will overload data centers, so they are announcing higher spending on dedicated infrastructure.

imgix · Unsplash License
- 0views
- 0likes and saves
- 0skips
In brief
- The text concerns companies' concerns that their infrastructure will not withstand the loads generated by artificial intelligence.
- According to a study, most enterprises will shift part of their investments toward managed services within two years.
- Two-thirds of companies expect moderate or significant shortages of accelerated computing power, such as GPUs.
The biggest infrastructure challenge
Companies increasingly believe that artificial intelligence will become a source of serious problems for their technical backend. Most enterprises will shift part of their investments toward managed service providers within the next two years (Firmy obawiają się, że zabraknie im mocy obliczeniowej, pamięci i storage’u dla AI – crn.pl). Decision-makers explain this by the fact that deploying AI may increase dependence on cloud providers, AI services and data centers.
Overloaded data centers
Respondents predict that AI-related workloads will overload both internal and external data centers within the next two years. The study covered decision-makers responsible for infrastructure and data at large international companies from the Americas and Europe.
Computing power shortages
The shortage of accelerated computing power, such as graphics processing units (GPUs), may be the most keenly felt. Two-thirds of enterprises expect moderate or significant shortages of this type of resource. Companies therefore anticipate that access to hardware will be limited and its cost will rise.
A response in the form of greater outlays
Consequently, enterprises intend to significantly increase spending on dedicated AI infrastructure. Some of them will choose managed services instead of building their own resources, which is meant to reduce the risk of overload and make their operations less dependent on hardware bottlenecks.
Read also
- Sztuczna inteligencja zmienia pracę. Eksperci wskazują na słaby punkt firm
- Polska chce mieć własną moc obliczeniową dla AI. Gate Cloud planuje centrum danych za ponad 13 mld zł
Sources
Discussion0
- No comments from this perspective yet. Be the first.