Interesting links, week 43, 2023, with some braindump
I was listening those podcasts with a lot of attention this week:
- Will Page interview in telecom.com: very nice pod on how Spotify started in Holland via clever bundling with KPN, and a good exposure to the music industry costs structure.
- Telco Super Ligero 145: [ES] Nice analysis of the poor state of 5G in South Korea. If south korea has failed, what is left for the telco industry?
Now if I had to play the cheap industry analyst:
- Mobile networks are mostly used outside premises (i.e.) between home and work, and they are used for media consumptions, but medias are limited by the resolution (small screen) and streaming technology side-effects (buffering). The content providers have a counter-incentive to minimize the intensive continuous bandwidth usage (i.e. faster/cheaper to send over from the CDN network.). As a result, bandwidth usage is not growing so much. Then, there is so much availability of eyeballs time. So there is stagnation in sight.
- Fixed Wireless Access is indeed something, but compared to a proper 5G service, fiber will be always cheaper to install/deploy on high density cases, and with low density, it means users are starving on the same peak hours if they don’t have enough sectors/bandwidth capacity. in 4G/5G NSA high capacity came from bandwidth aggregation, which is by definition less doable in rural area, so 5G SA is a better option there, but requires a new rollout, and will be less cost effective if “normal” phones used in rural area are not using it (read LTE only phones). Classic chicken-egg problem at each generation, but exacerbated.
- New RAN vendors (OpenRAN!) were doomed to fail: to enter a new market, you need to build a simple, disruptive enough (pricing here) product. However, when TIP started, the industry engaged itself in a uber-complex architecture as first MVP. Using the Intel stack originally used for FlexRAN, the industry (tier 1 + triad RAN vendors) convinced new entering RAN vendors should engage directly into solving the DU/CU architecture model which makes sense in high density scenario, i.e. your macros on Time Square or plaza Catalunya. This also means you need not only to design a complex product on day one, but run it in a distributed fashion, and mix it with realtime constraints. How could CTOs offices even think new vendors could remotely succeed? That was probably never the objective. Anyway.
From the Super Ligero podcast we learned that South Korea, which has probably the most advanced usage on the planet in terms of media/technology, has not been able to have new devices/usage pairing that could justify the need for the amazing bandwidth capacity in 5G SA on high bands. Immersive handset like the Apple one could have a role, but as we see from Apple other devices itself, the progress in custom silicon, combined with on-device AI processing, makes the network capacity itself less and less relevant for the new generation of personal compute.
So what could be the kicker for telco industry? Focusing back on cost effective coverage.
Existing/new actors might have a role to play in coverage gaps filling if the radios are cheap enough (i.e. use a lot of tiny ones) and if the coverage provided is bringing value (i.e. lot of roaming users doing traffic/calls). However:
- Because of the obsession of fixing macros, the smallcell market has been completely left abandoned (I play with the same actors in my successive work labs since 2017). We deal with the same great ODMs building small radios, with OEM executing quite well (i.e. like Baicell), but the demand is still very limited in units. In Spain for example, we have 33k macros cells deployed across 4 operators according to antenasmoviles.es, which means if smallcells were to be used to extend the coverage in specific dead spots, the market size would still be limited compared to the WiFi APs market (which is one per home at minimum). Pico LTE cells (20mw, sub 1W) are still 5x the price point of a WiFi AP for a 1/5 of the bandwidth (i.e. a $100 Wifi6 AP is doing like 700Mbps in my crowed wifi area, my TDD lab cell will be forever capped to 100Mbps at 20Mhz).
- Private Networks is not taking off due to the niche market problem, so unlikely to bring a demand for a lot of units. And customers that want/need a private network probably can afford a real RAN deployment (i.e. mines/oil/port/warehouse industry) which is de facto making the entry of new small actors very difficult (i.e. expectation on quality, resilience etc, pretty high for a first gen product).
- the decentralized wireless (Dewi) folks have entered the space with cryptos reward schemes that remind me what FON intended in 2006 with Wifi access points used to distribute connectivity. But this line went to bust and they survived because they partnered with tier 1s into offering WiFi roaming on the provider’s wifi boxes. This time, Dewi are however bringing SIM roaming as part of the deal (great!), but then it is limited to country that have a light framework for Private Networks - such as US with CBRS, and private network spectrum chunks across the planet are still not aligned globally, limiting the interest of potential radio vendors due to lack of units - need a RF frontend per region. Who is gonna burn money on those coverage extension in such a way ?
But the telco industry is bouncing back already. Everything seems to point at that big telco actors will instead engage with another industry to solve their coverage need - Space!. A lot of signals are driving the telco industry to this old holy grail:
- cheaper launch/reuse rockets. Demonstrated viability of the LEO model via SpaceX and Starlink, and capability to iterate quickly on new satellite tech by riding on the next week launch.
- working technologies such at Emergency SOS from Apple saving life with space comms on COTS handsets with the proper band.
- AST SpaceMobile PoC seems to deliver those comms capabilities across standard handsets in some operator trials.
- NTN getting quickly standardized in 3GPP bodies.
- tier 1 operators will persist into the thinking this is more cost effective than bringing coverage on the ground in many geographies, despite the high environmental costs of space launches and light pollution of LEOs.
- a lot of challenges with the interactions with the ground network (i.e. interferences, roaming, handovers capabilities etc) in the next iteration of antenna in the sky tech, especially for data usage, but promising as engineering challenges.
- enable industry to recover from the 5G fiasco and focus on delivering an actual value for customer.
Personally, I will be watching that space (pun intended) in the coming months/years.