We introduced the Private 5G “readiness wheel” last year, to illustrate the maturity of the private 5G ecosystem or lack thereof. The “wheel” isn’t rounded yet, so it highlights the challenges of scaling the private wireless market.
In our newly released Industrial Private Cellular market report, we show the current state of the wheel and highlight the changed macro-environment that will likely result in steady – not exponential – growth.
One new factor to consider is the craze around artificial intelligence, specifically generative AI (GenAI). With the introduction of OpenAI’s ChatGPT-4 last year, GenAI has captured the mindshare of the general public and businesses.
According to this article, ChatGPT reached 100 million users just two months after its launch. Unlike traditional AI/ML, which is good at data analysis and pattern recognition based on existing data, GenAI uses machine learning models to learn and create new data and content.
This ability to create and improve new data content enables GenAI to “reason across audio, vision, and text in real time,” according to the announcement of OpenAI’s latest “omni” GPT-4o. With genuine excitement about the potential uses of GenAI in business – e.g., new drug discoveries in healthcare, new compound material discovery for renewable energy, replacing customer service agents, and many more possibilities – businesses across almost all industries are exploring ways to leverage AI.
Lost hyperscaler leadership
AI investments are primarily confined to within the four walls of hyperscalers’ core regional mega data centers today. With each rack of AI servers with Nvidia GPUs costing $1 million+, not many businesses can afford to invest heavily in AI. Hyperscale cloud providers have a global scale and multiple sources of revenue to make that kind of “at-scale” investment.
The current flurry of AI infrastructure investments has distracted the hyperscalers away from 5G networking. With an extensive network of relationships with developers, partners and customers in the enterprise IT space, Microsoft and other hyperscalers could have played a vital role in scaling the private networking ecosystem. That leadership is lost at the moment.
System integrators can pick up the slack. We see the private 5G ecosystem progressing with telco-centric and enterprise IT and OT-centric suppliers introducing turnkey and discrete point solutions across core, RAN, LAN/WAN transport, devices and a roadmap towards unified network management/service delivery platform. It should be remembered that the private wireless market is growing at a healthy double-digit growth rate despite some dour headlines in the press in recent months… and it will reach $5.8 billion in infrastructure and device equipment sales by 2029.
Today’s market isn’t big enough yet to support too many players. It is reminiscent of a crowded pond with too many frogs muddying up the water. Some are pulling back; others are focusing on specific sectors. This is a natural business cycle of market forces weaning out winners and losers. But the pond is filling up with water, becoming a lake, and the fastest frogs will thrive.
So, is the current craze around GenAI an accelerant or inhibitor to private 5G networking?
Despite some interesting use of AI in the mobile network itself, we view GenAI as a headwind to private 5G in the near term. We see a mindshare shift to AI among major enterprises and key stakeholders in the enterprise IT ecosystem, including IT hardware vendors, system integrators, distributors, etc. Companies are diverting capital expenditure and resources towards AI projects and initiatives. However, we believe that future AI/ML applications and workloads will eventually move away from hyperscalers’ core data centers to the edge of enterprises.
There are plenty of data sets from sensors and workers in industrial and enterprise settings that require low latency and reliable connectivity links that private 5G networks promise. There are plenty of existing data sets that traditional AI/ML can process to gain insights for operational efficiency and worker safety before we take advantage of GenAI to gain newer insights and further productivity gains.
Over time, that “$1 million+” rack of AI servers will become a few hundred-thousand-dollar rack to be deployed at enterprise edge for edge computing. By then, we see enterprise private networks with 5G and other technologies providing “always-on” connections to myriad data sources to enable the AI future that businesses envision.
In the future, AI/ML will become a driver for private 5G, not a distraction.
Gen-AI may be sucking the oxygen out of the room for now, but in the longer term, we see the AI world connected with private 5G.
Original article can be seen at: