Welcome to the 23rd edition of the Beyond the Next Billion newsletter. In this edition, we take the opportunity to share some of our key takeaways from the week that was, so to speak, in Barcelona at Mobile World Congress 2024. We are working on a more detailed version of the summary below which will be available to clients. DM if interested.
The week that was - general thoughts
“Business as usual” was the very first takeaway from the week that was in Barcelona, in the sense that there were a lot of people at MWC24 and a lot of the obligatory fantastical concepts like the flying taxi demos on display this year (courtesy #e& and #SKTelecom). The GSMA’s official attendance count was pegged at 101,000, and it certainly felt like it, with throngs of people jostling for space in the main hallways and thoroughfares. Having arrived in Barcelona on the Friday preceding the show, and jumping straight into meetings and catch ups, MWC24 got off to a pretty hectic start. Needless to say, by Wednesday afternoon, we were looking for the quickest shortcuts to get to meetings and also ducking into the relative serenity of the Media/Analyst Center whenever possible for a breather.
“Bigger than usual” was the second takeaway about MWC24. While most of the usual suspects were in their traditional locations in the various halls, there were a few changes and some new configurations that threw us off a couple of times, especially when hunting for the Country Pavilions. Anyone take a leisurely walk out to Hall 8? Admittedly, we rarely get out to Hall 8 but this year, many of the special events and summits were housed in Hall 8, which seemed even further than in previous years, and also more cavernous (new space added?). To be clear, we are not complaining about MWC getting even bigger, rather more a commentary on how things seem back to an almost pre-pandemic normal.
MWC 2024 - Key takeaways
Just prior to the show, we had put out a note with our “key expected themes” to watch out for. It would be fair to review those expectations against what we actually saw at the show. Mercifully, we believe we got a number of things right but there were surprises at the show.
Artificial Intelligence
AI, as predicted, was the buzzword of MWC24. “AI-enabled”, “embedding AI”, and “GenAI” were some of the most commonly heard terms during virtually every meeting and across most of the vendor booths. We had discussions about AI being utilized in the radio access network (RAN) for optimizing resources, AI on devices (the show was abuzz with new AI-enabled devices from Samsung and more), as well as GenAI for telcos.
AI in the RAN and in the telco network is not a new theme but there is clear progress being made on this front. A major factor in the increased importance being given to using AI in the network is the very real and tangible concerns that operators have that the 5G+ future will mean new densified architectures and soaring energy costs under the current paradigm. On the device front, we are somewhat skeptical about whether AI will really move the needle for the smartphone industry but this is a question best left to the device analysts. GenAI came up in several conversations that we had with the likes of #Amdocs, #ServiceNow and others.
It is increasingly clear that there are a number of valid use cases for GenAI within a telco environment but at the moment, the most obvious ones continue to be within customer-facing functions like customer support, contact center, field services and others. While there are definitely opportunities to “optimize” costs, the real value and ultimate hope will come from true “AIOps”, where telco networks, built on a modern data architecture, can realize the promised land of real digital transformation.
But, this poses a number of questions.
One, what kind of #LLMs do telcos need to be able to enable meaningful GenAI use cases?
Two, should these be based on general purpose GenAI platforms like #OpenAI from #Microsoft and/or #Gemini from #GoogleCloud, or should these be built with domain-specific models? Players like Google Cloud seem to be suggesting that general purpose LLMs would do fine with some tweaking, while ServiceNow seems to favor a more domain-specific approach.
Three, can these LLMs be built on top of existing data architectures within telcos without soaring costs for data transfer and processing? Or do they need to upgrade to the more “modern” architectures offered by the likes of #Cloudera?
We believe that currently, there is simply too much proprietary data within the telco domains to move the needle with general purpose LLMs. In other words, there is a reason that GenAI can only help in the areas mentioned above, because they can pull from relatively standard IT platforms. The network domains are too complicated at present. However, if GenAI is to extend across the telco domains, then there needs to be an interim stage of upgrades to a modern data architecture. AI requires a lot of compute, as well as real-time data, both of which are difficult to do with today’s network systems.
There is also an interesting trend to watch with telcos pitching their credentials as providers of “Sovereign AI”. Singtel, the incumbent operator in Singapore, had an announcement with NVIDIA in early February about deploying the latter’s full-stack AI platform on their Nxera data centers in Singapore. The NVIDIA blog on this announcement calls out the recent movement towards Sovereign AI, where telcos essentially build “AI factories” that can comply with data sovereignty regulations. These would build on country-specific LLMs that are necessary in many regions of the world to cater to local languages, dialects and nuances that are simply not possible in today’s construct. It’s still early days for AI, but there are very real concerns in this new era of geopolitical dissonance about depending entirely on “foreign” technologies for AI. Telcos stand to gain significantly by tapping into these trends and offering compute and compliance for AI within individual countries.
Private 5G
In the pre-show note, we had commented that our early enthusiasm about private networks was now 5 years old, and was at risk of being consigned to the analyst “sin bin” of predictions that were either plain wrong or never panned out. MWC24 left us with enough evidence to suggest that #Private5G is not quite done yet. If anything, it was heartening to see the healthy doses of realism that seem to have seeped into the industry. The fundamental challenges of spectrum access and business models still persist and continue to constrain large scale rollouts. Private 5G, despite many similarities in architecture and commercials, is largely a local affair in the sense that specific countries’ policies will largely determine momentum for private 5G. Last year, we had made some of these points in our post-show roundup, which you can read here.
That said, a number of things have changed for the better, and in some ways, it felt like this was an industry that is quickly “maturing”. Of course, it would be fair to ask how could an industry segment that has yet to take off be in a mature stage? We only mean that industry attention has shifted to filling the gaps in the private 5G proposition. In 2024, private 5G narratives will continue, with multiple vendors doubling down on their investments. But the value points are shifting away from the radios and mobile cores, which are now table stakes, towards the distributed edge and seamless orchestration of workloads across enterprise IT/OT environments. We explain below with some additional detail:
One size doesn’t fit all - Enterprises need a lot of options when it comes to their requirements. This has been a challenge in the past, where the more “robust” solutions from the major equipment vendors had been prohibitively expensive (Ericsson in particular but even Nokia to some extent), while the more “affordable” solutions from the likes of Celona, Athonet et al left open doubts about their effectiveness and reliability at scale. The landscape is much improved at the start of 2024, with options galore. Nokia made an important announcement of a Compact Digital Automation Cloud (DAC), in partnership with Intel. As the market leader, the Nokia dalliance with Intel and its acceptance of FlexRAN architecture is a subtle shift in strategy, an acknowledgement of the fragmented nature of the private 5G opportunity and the very long tail that represents the majority of the market.
Distributed edge takes center stage - to enterprises, taking only a bunch of radios and a mobile core of their choice is an incomplete solution with poor ROI. The real value in private 5G was always going to be in the edge solution that gave enterprises the flexibility to deploy workloads and handle processing on-premises, or segregate some workloads to be processed through multi-cloud connectivity to public or private clouds. RedHat made a compelling argument for using OpenShift at the “far edge” with an interesting announcement with Intel and Minsait to utilize FlexRAN and OpenShift for industrial workloads. #Arrcus had a really interesting announcement with #Softbank, the Japanese operator, to deploy 5G multi-access edge computing (MEC) by using SRv6 Mobile User Plane (MUP) to simplify edge compute and workload transfers with low latency.
Orchestration of workloads needs to be seamless - this is an area that has always been taken for granted in private 5G but is assuming more importance, especially as enterprises have started demanding on-premises and multi-cloud connectivity and easy movement of workloads. To be fair, there are already a number of solutions that claim to handle orchestration in the way that service management and orchestration (SMO) solutions do. We had an interesting conversation with the CEO of #AarnaNetworks that offers such an orchestration solution called #AMCOP, validated with #DruidSoftware and #Airspan, that is seeking to address this exact requirement.
Device and infrastructure security is increasingly critical - industrial devices that plug into private networks have traditionally been a big challenge but the availability has improved a lot in recent months with ruggedized and specialized devices becoming available. Securing these devices has been talked up but 2024 looks like the year that real progress is made in this area. We had two conversations on this front. The first was with #OneLayer, an Israeli company specializing in private network security, which has a unique approach to enabling full visibility into the devices that are plugged into the private network, including those sitting behind routers. The second was with #Nexcom, a Taiwanese company that has introduced a “security router”, for lack of a better word, that has been deployed by a Japanese industrial giant to secure all of their robotics on site.
Systems integrators (SIs) continue to crowd into this space - this is a point we had made in last year’s post-MWC roundup note on Private 5G, with the global consulting giants, Indian SIs as well as domain and country-specific SIs all jostling for a piece of this business. Country-specific SI’s will be a big factor in APAC in markets like Japan, Korea and Taiwan, where local regulations play a big part. At MWC24, there were some new entrants into this crowded space. One of the companies we spoke with was #Amdocs, which has launched a new proposition for private 5G. Amdocs showcased its “Network in a Box” solution, which has integrated with Intel’s FlexRAN software stack, testing and validation, and also added a management layer to simplify deployments for enterprises.
Fixed Wireless Access (FWA)
In the 2023 post-show note on FWA (the 4th in a long series last year), we had identified three emerging segments for FWA:
Sub-6 GHz with 5G
mmWave E and V bands
mmWave 24-38 GHz
Of these, sub-6 GHz 5G is the runaway winner with virtually all of the current deployed installed base in these bands. It is no small coincidence that sub-6 GHz became the band of choice for FWA, as it also mapped to what quickly became the default global band given chipset and device availability at the time.
The biggest markets for FWA continue to be the US and to a lesser extent, Japan, with virtually all connections on sub-6. But there is a new kid on the block, and one that should generate significant volumes, which is India. In a recent Mandala Insights report (DM us for details), we had projected that India would see 36.6 million connections by 2028, with the majority of connections on sub-6 GHz but a solid number of mmWave as well. One of the biggest challenges for markets like India has been the affordability of outdoor units (ODUs) but incumbent operators like Jio and Airtel are coming up with innovative solutions to connect multiple homes from a single deployed ODU. These designs seek to reduce the OPEX burden for telcos by reducing the number of truck rolls that are needed for installation while also trying to balance available bandwidth per household.
As we had predicted pre-show, mmWave was prominently visible in a number of booths, with CPE designs being shown and some demos as well. #Airspan was running a prominent demo at their booth but with no representative available at the time to engage with. Despite plenty of visibility for mmWave, what is less clear is actual availability and timelines. When pressed, one common response we received on a few occasions could be summarized as “whenever #Qualcomm gives us the chips.” Before this gets misconstrued, I should clarify that clearly, not all CPE vendors have multiple chipset suppliers so this is not representative of all CPE vendors. However, it would suggest that the market continues to wait for mass availability of affordable mmWave solutions but the same comments left enough room open to suggest that this problem could be resolved in the near future.
mmWave availability, in our view, can not come soon enough. We have asserted in the past and stand by our hypothesis that eventually, mmWave becomes the “overlay” network for FWA as sub-6 starts to get clogged with consumer MBB traffic. We also tested our hypothesis against recent claims of operators deploying network slices for FWA, but conversations with equipment majors like Ericsson confirm that this is only meaningfully possible today with standalone networks. But the proportion of SA 5G is still relatively small, so sub-6 FWA will run out of capacity at some point in the near future.
Some of the FWA announcements at MWC24 that caught our eye include:
Nokia’s new mmWave FWA outdoor receiver. While we were briefed about this prior to MWC, Nokia did announce this officially at MWC24, and this builds on their pioneering work to get mmWave CPEs enabled for the mass market. This new product will boast 27 dBi antenna gain for extended range, as well as using advanced Bell Labs techniques to enable a 360 degree field of view for catching the strongest signals.
Qualcomm’s new FWA product. The FWA Gen 3 platform should offer a number of enhancements for the mmWave proposition. Based on the Snapdragon X75 modem, it supports dual connectivity for sub-6 and mmWave, as well as tri-band WiFi 7. But what was more interesting was the “under the hood” showcase of the motorized reflector with a 360 degree field of view. The idea seems to be to utilize AI to help the mmWave antennas in the 3rd gen QTM567 module to tune to the strongest signal source. No information was available on sampling so here’s hoping this hits the market fast and at scale.
Airgain’s Smart Lantern FWA. The product looks like a CPE but the secret sauce as claimed by Airgain seems to be their smart beamforming technology that identifies the strongest source of signal automatically with little to no intervention by the customer. While Airgain is not the first to claim a smart antenna design, there is every chance that they have built a better mousetrap. Airgain also has a couple of other interesting products, including an interesting “roof-mounted 5G gateway” called AirgainConnect, which is installed on the roofs of buses and vehicles to enable WiFi access on board. While this is not FWA strictly speaking, it is a really interesting extension of the use case.
BEC’s AirConnect. Our very first meeting at MWC was with BEC who we have spoken to before. This Taiwanese company has a number of 5G CPE designs in the field already, for both indoor and outdoor scenarios. The 8235 outdoor CPE looks like a really solid product combining sub-6 5G and 4G LTE. BEC also had a number of products for industrial and indoor environments.
Ericsson’s Cradlepoint subsidiary announced an enterprise-focused FWA product. While we did not get to see this product live at the show, we have always believed that FWA would be relevant for the enterprise, especially the small and medium business (SMB) segment. SMBs would benefit significantly from an industrial grade FWA device that also couples with a cloud manager, which is exactly what the Cradlepoint X10 solution seems to be targeting. This is an interesting space to watch.
Sustainability in the network
This is one of our pre-show predictions that we must confess was a relative miss. There is no question that sustainability is a hot topic within the telecom industry but at MWC24, it seemed that sustainability messages were embedded into overarching messaging but it was not quite the centerpiece. This was somewhat curious given that overall industry awareness of sustainability concerns are now quite widespread. Or perhaps, this is exactly why it wasn’t a buzzword at MWC24 but rather, the industry has already moved into an execution phase.
That said, there were a couple of interesting things we would like to highlight from the show:
We were lucky to be invited to moderate a Fireside Chat with EDX Wireless and Ericsson on the topic on February 27th. This was an engaging session, and zeroed in on the RAN as the lowest hanging fruit to tackle in terms of reductions in energy consumption. However, as operators continue to deploy and expand 5G commercially, a lot can be achieved from the beginning if sustainability features are incorporated into the planning tools. Beyond incorporating weather patterns into these tools, network engineers can benefit from the availability of digital twins that simulate energy consumption across multiple scenarios. Vendors like #EDXWireless are working on these kinds of technologies to add to their existing portfolios.
We had a very interesting meeting with #CombaTelecom, a wireless infrastructure company that specializes in small cells, DAS, cellular base stations and other equipment. We were given a sneak peek at their #Helifeed 2.0 Green Antenna. This product has a highly modular design that has eliminated a number of components to reduce the total bill of materials and as a result, energy consumption.
There are several other things we haven’t been able to touch upon in this edition in detail, including telco APIs, 5G monetization and more. But we do hope to cover these in upcoming editions. As always, do reach out to share feedback and DM with any queries.