As AI agents grow more autonomous and take on increasingly sophisticated tasks, some tech companies suggest that computing might start to shift closer to the user.
Cisco President Jeetu Patel refers to this developing model as “desk-side computing”—a framework where dedicated AI devices positioned next to users collaborate with cloud infrastructure to manage AI tasks.
What is desk-side computing?
Desk-side computing involves having a specialized AI machine located near the user rather than depending solely on remote cloud servers.
These devices could include compact computers like Apple’s Mac Mini and Mac Studio, or specialized systems powered by Nvidia or AMD processors. They would consistently run AI models and agents, functioning almost like a personal AI workforce.
Although “desk-side computing” is not yet an established industry term, it symbolizes a larger trend towards edge AI and local inference, where certain AI tasks are processed closer to the user, rather than entirely in the cloud.
Why are AI agents creating new computing demands?
Conventional computing was designed for human users, who make requests sporadically.
In contrast, AI agents can continually reason, search for data, engage with models, and perform tasks without breaks. Multiple agents might operate simultaneously, leading to persistent workloads and generating significantly more network traffic than a human user would.
Patel notes that as AI agents enhance their capabilities, they will constantly share information and update their memory, increasing the demands on processors, storage, and networking systems.
This could necessitate far greater computing power than what today’s personal computers are equipped to deliver.
Why might some AI workloads move closer to users?
While cloud infrastructure is not likely to vanish, with large language models and intricate AI training remaining reliant on massive data centers, certain tasks may increasingly be processed locally.
Bringing AI workload processing closer to users can lower latency, enhance response times, and offer dedicated computing resources without complete reliance on remote servers. Local processing might also help decrease cloud costs and enhance privacy for specific applications.
Desk-side computing doesn’t aim to replace the cloud but suggests a hybrid model where workloads are divided between data centers and local devices.
Why does this matter for networking?
This transition could have effects that extend well beyond personal computing.
If employees are aided by numerous AI agents functioning continuously, the demand for network bandwidth, processing power, and energy could surge dramatically.
Businesses may need to revamp workplace infrastructure to accommodate this new computing paradigm. Enhanced networks, advanced chips, and improved data management could become essential.
For Cisco, a provider of networking equipment and infrastructure, this represents a promising long-term opportunity.
Why does Cisco see desk-side computing as the next phase of AI?
Patel envisions a future where AI agents serve as digital collaborators, managing research, coding, analysis, and administrative responsibilities while humans focus on oversight and decision-making.
In this scenario, an individual could potentially supervise numerous AI agents at once. Accommodating such workloads may necessitate dedicated AI machines that function alongside traditional PCs and cloud services.
If this vision comes to fruition, the personal computer could transform from a device for one user into a platform that manages an entire team of AI workers.
What could prevent this shift?
This notion remains more of a vision than a recognized industry standard.
Specialized AI hardware might be costly, consume more energy, and introduce complexity into enterprise IT infrastructures. Many workloads could still be performed more efficiently in the cloud, leading companies to prefer hybrid architectures over a fully decentralized setup.
Nevertheless, as AI agents become increasingly autonomous and demand more computational resources, the balance between cloud computing and local processing could inevitably change.
That’s why Patel believes desk-side computing might signify the next substantial shift in the evolution of AI.
