A global survey of over 850 leaders of artificial intelligence (AI) initiatives conducted by The Futurum Group finds more than a quarter report their organization is wrestling with a skills shortage, with other challenges including legacy system integration (35 percent), IT resource constraints (32 percent), complex AI technology stacks (30 percent) and data quality and governance concerns (27 percent).
Spending boosts MSP opportunities
Those issues, naturally, bode well for managed service providers (MSPs) and consulting firms. A previous Futurum Group survey of over 1,000 business and IT leaders involved in AI application initiatives found that 73 percent of organizations plan to change or add new consultants or system integrators in 2025. That same report also noted that 61 percent of organizations are already relying on outsourced AI solutions.
Less clear is where AI applications will ultimately be deployed. Consumption of AI services in the cloud is on a per-token basis, with each input and output requiring a separate token. IT organizations are quickly determining that the cost of tokens for inputs and outputs when relying on cloud service providers quickly adds up. In fact, given the amount of data required to drive AI applications, on-premises IT environments are proving to be a more economical option for deploying AI inference engines. Add on top of that compliance and performance requirements, and a very large percentage of AI applications will be running in an on-premises IT environment.
That doesn’t mean the cloud won’t play a critical role in training, customizing, and experimenting with AI models, but it does mean there is likely to be an on-premises data center resurgence in the age of AI. In fact, research from The Futurum Group finds that 69 percent of respondents work for organizations planning to change or add new AI server vendors in 2025. Currently, Dell (49 percent), IBM (45 percent), Cisco (45 percent), and Oracle (44 percent) are the top choices.
MSPs and the AI data shift
The irony of all this, of course, is that many organizations have abandoned data centers in favor of cloud services. Now, many of those organizations are once again looking to either build data centers or rent space in a colocation facility. The challenge with the former issue is that many organizations no longer have the expertise required to build, much less manage, a data center. As for colocation facilities, vacancy rates are currently at an all-time low, so there might not be that many options for running AI workloads in an on-premises IT environment.
Inevitably, organizations will look to MSPs with AI expertise to help them solve these issues. Exactly how MSPs have the expertise needed to successfully deploy, manage, and secure AI applications is unknown, but the demand already far exceeds the available supply of MSP expertise.
Of course, the channel, much like nature, abhors a vacuum. IT vendors have already made a host of AI training to help drive the adoption of managed AI services. The only thing that remains to be seen is how soon MSPs make the most of that opportunity.
Photo: David P Baileys / Shutterstock
This post originally appeared on Smarter MSP.