In his final earnings call as Ericsson CEO, Börje Ekholm said that the vendor will be raising prices on new tenders and also re-negotiating price rises into existing contracts as the impact of inflationary component costs threatens to take a toll on the company’s margins.
Ericsson does not include automatic pass-on of component price increases in the vast majority of its contracts, so any contract changes need to come about as part of a renegotiation with customers. Ekholm said that the vendor had already successfully re-negotiated some contracts, as customers understood that they needed to work with Ericsson to share the load of the rise in semiconductor prices, driven by the AI boom.
These aren’t easy discussions, he admitted, but he also said that customers do understand Ericsson’s position. “When we enter into these discussions we need be thoughtful but where we have done it we are seeing customers understand that we need to find ways to share burden of the industry if it will be competitive going forward. We have the opportunity ahead of us to do more,” Ekholm said.
“In the longer term we cannot take it [component inflation] all alone – it is what we can do with customers to ensure we get the best performing solutions to customers but also at right price point.”
The company has also been mitigating the impact of component price increases by substituting products, and redesigning products “in a way that minimises cost inflation,” Ekholm said.
He added that he thought that all vendors, especially those that cannot sell products that contain potentially cheaper Chinese components into certain markets, would have to go through the same processes.
Ekholm farewell, hopes for physical Ai
In wider comments as he handed over to incoming CEO Per Navinger, the departing Ekholm said he was proud of the progress Ericsson had made since he took over in 2017 to strengthen its technology leadership and improve operational execution. The company, he said, is well positioned for long term success.
Departing CEO Ekholm handed over to his replacement Per Navinger during the Q2 results call.
Ekholm said several times that he believed the AI boom is about to move into a phase where high performance mobile connectivity is essential, especially in the uplink. Ericsson is well positioned to benefit from this, he said. “I’m a big believer AI will move into the physical world and when that happens we will be very well positioned with the initiatives we have taken,” Ekholm said.
“Our strategy is focussed to position us for the next phase of AI adoption. When it moves into the industrial or physical world, connectivity will be more important and uplink will dimension mobile networks. We will also see increasing demand for low latency and this is what 5G was designed for. Ericsson is well positioned to capture the next phase of AI connectivity.”
Defending Ericsson’s decision not to go after the datacenter boom, he said, “We have elected to be in a different part of the value chain of AI. I think the next phase of AI is going to benefit our industry quite substantially. For sure we could have done it differently, but we have also done what we can to position the strength of Ericsson in the best possible way to where market will be in the future. And we are convinced we will see AI move into distributed apps – anything from glasses, humanoids, robots – and when start to see that it will demand high performance mobile connectivity with solid indoor coverage and high uplink. That is where we have invested.”
No immediate AI boom for RAN sales
That said, he doesn’t see this phase of physical AI driving any sort of near term boom in RAN investment. Ekholm said that the RAN market, still the core money maker for Ericsson, is likely to remain “flattish” for the foreseeable future.
“We are starting to see emerging demand for uplink but we can’t really pont to exactly what type of applications that involves. In a broad-based way we are starting to see demand for AI starting to shape traffic. It is an upside case to start to be a bit more optimistic about our industry and the RAN market.
“At the same time, for our own planning, we need the right products and cost structure and we need to be disciplined in the way we execute and plan our cost structure. So… cautious but in a few years time these purchase decisions will be in the hands of our customers and when they see demand happening they will start to buy. But until then let’s continue to plan for a flattish market.”
The company, he said, will continue to look for growth areas where it can leverage its core strengths.
One area the company has targeted is in enterprise, although this has been a tough road to date. Although the enterprise market is still making a loss, Ekholm believes that the company can turn that around, with some encouraging signs from Wireless WAN. However, enterprise private networks has not had attractive growth or profitability, although the CEO claimed the company is working on it and starting to see progress. Similarly, although the Vonage acquisition has also not been contributing and the company has taken large write-downs, there is a plan in place to change trajectory which will take a bit more time andthe company is starting to see improved performance in reported numbers, Ekholm said.
What will be critical for long term value creation will be to find new uses for Ericsson’s technology. Here, mission critical and defence applications are also already contributing, he said.
“What we have been focussing on is to drive growth into the company without being bluesky: be disciplined in way we enter areas, invest to capture potential, and to get into growth. That’s where the next step of the journey is.”
GPUs and AI in the RAN
As for the future product direction of its main RAN workhorse, Ekholm stuck to Ericsson’s line that it is “agnostic” on the underlying hardware platform.
He said that Nvidia’s entry into the market is confirmation of the importance of AI in the RAN. Here, Ericsson has already taken steps to enable AI uses cases in its radios, using its own dedicated silicon and compute.
“It confirms what we have been talking about – that AI will be what drives networks going forward. We have picked a strategy of being agnostic from a hardware point of view. We can run our RAN stack on x86 or GPU or our purpose built silicon.
“When we look at what you need in the radio – very high performance, energy efficiency and a lot of calculations – it is a demanding compute environment. At the same time we do not see a need for very large models, so where this market ends up is always a bit uncertain.”
So that compute demand can be met by Ericsson’s own purpose built Radio Silicon, but the RAN stack is agnostic and can be on whatever type of infrastructure wins, the CEO said.
“It’s not either/ or. We are saying let’s see where the market shapes up. Today there are clear benefits in cost, energy and performance from purpose-built, and how it looks over time we are not placing bets, we are simply keeping it an open topic.”
