Based on the presentations I linked, Arista and Cisco are investing in Linear Pluggable Optics (LPO) solutions instead of CPO. Nokia is not known to be developing its own LPO products, and the rationale may be the following (GTP5.2):
"Linear pluggable optics is a real and rational competitor to CPO in the short and medium term, and it is already in limited production use by hyperscalers. It offers a large portion of CPO’s energy efficiency benefits without the risks of co-packaging. However, LPO does not eliminate the port density and electrical limitations of the pluggable architecture, so at the highest speeds (1.6T+) and scales, CPO is inevitable in the long term. LPO and CPO are not alternatives but successive steps.
Nokia does not appear to be developing LPO as its own strategic optics product, but rather treats it as a transitional solution that it supports at the system level, but does not build its long-term competitive advantage on it. Nokia’s actual investment is in CPO and optical engines (InP + SiPh), where it can transfer architectural value-add to itself. LPO is a compatibility and customer path for Nokia, not a destination."
Here’s some writing about those optical modules (pluggables) which are built for inter-data center connections. This is already in use - and it’s driving Nokia’s current optical business growth. (Scale-Out)
And CPO (Co-Packaged Optics) is intended for inside data centers. And presumably, this construction will begin this year. (Scale-Up)
Optical Network Transformation in the AI Era
The optical networking industry is undergoing its most significant transformation in decades, accelerated by the immense bandwidth demand generated by artificial intelligence (AI) (forecast 913 ZB/s in 2028, 30 times the current). This demand places unprecedented pressure on network power consumption and physical space, as data center electricity consumption is projected to double by 2030.
1. The Coherent Module Revolution (800G ZR/ZR+)
Coherent pluggable optical modules have met this challenge by offering higher performance at a fraction of the power. Power consumption per gigabit (W/G) has decreased by 99% since 2000 to 0.03 W/G with the latest 3 nm CMOS technology.
New Standard:800G ZR/ZR+ modules are the new standard, utilizing 3 nm CMOS technology. They extend the application range from data center interconnects (DCI) to metro and backbone networks and the connection of AI clusters.
Versatility and Flexibility: For example, Nokia’s ICE-X 800G ZR+ can achieve a range of over 1,700 km at 800 Gb/s and up to 3,000 km at 400 Gb/s.
Maximizing Investment (ROI): The programmability and backward compatibility of 800G modules (e.g., connectivity to 400G routers) enable gradual network upgrades without the need to prematurely replace expensive routers, thus maximizing customers’ ROI.
2. The Next Wave: 1600G
The next step in development is 1600G ZR/ZR+ modules. This generation of solutions requires 3 nm and sub-3 nm (e.g., 2 nm) DSP chips, more complex two-subcarrier architectures, and the use of advanced materials (such as silicon photonics, InP, and thin-film lithium niobate) in even tighter integration.
In summary: Coherent pluggable optical modules are the cornerstone of the optical industry’s transformation, as they offer the best way to support the exponential bandwidth needs of the AI era by reducing power and complexity. 800G is the current standard, and development towards 1600G is already underway.
Hotard is at least doing things. Somehow Elop comes to mind, perhaps a slightly better version though.
The HR director’s departure in the summer might at least partly be due to this. In practice, this is where those who will be laid off in the next change negotiations are chosen.
It may well be that there are units where there simply isn’t a single top or terrible employee, but I don’t personally believe in such a thing.
I think that with this move, Hotard will now see the reason why productivity development in Europe is so far behind the US. You can start looking from the answers of the shop steward and professors.
I guess they are trying to wake people up here, a bit like Elop woke people up with his “burning platform” email. The risk might be that the aging workforce in Europe turns against the management. What makes selecting underperformers difficult is that Nokia has had co-determination/change negotiations (YT-neuvottelut) practically every year for the past 20 years, and in these, most clear underperformers have typically left. If a team of 10 still has two underperformers in 2025 and they leave in the next co-determination round, then next year, in 2026, new underperformers would have to be found from the same team, and so on.
It’s kind of Taylorism, sometimes dowadays not good!
Central to Taylorism is the transfer of control over how work gets done from workers to management. In the 21st century one version of that thinking, Digital Taylorism, is evident at companies that track workers’ computers, phones and other technology to be sure the company is getting the most out of its employees.
Well, well. I think I’m about to stir up a ‘hornet’s nest’ now. I’m thinking about Satonen’s new fancy ‘kick law’ versus this categorization of Nokia employees. It seems like 5% will leave around January 2026 with a small shrug of the shoulders. It’s going great.
A 5% dismissal does not require lowering the bar for personal grounds for dismissal. As before, company management can simply state that the organization wants to focus on other matters in the future, or even save money, and this can be achieved by reducing the workforce. Change negotiations are held, measures are taken, and after the notice periods, 5% of the company’s workforce is gone.
Nokia has had an employee classification system for decades, and its significance for employees is great because it affects bonus multipliers. In this respect, there is nothing new. When the model was originally introduced, it was similar to the JH model that has now come to light, meaning teams were forced to name the “weakest” segment, cut their bonuses, and draw up a development plan, the implementation of which was monitored unless the person resigned/was dismissed before its completion. Quite quickly, the rules were relaxed, and the classification mainly rewarded meritorious employees. An “Improvement needed” assessment was possible, but it was rarely given, with a small bonus impact and in an encouraging spirit.
It seems really bad that Nokia’s HR issues are being reported in this way, a major reputational damage. It also doesn’t sound good that the new HR director appointed by Hotard has adopted a forced distribution model for personnel evaluation based on the 1980s rank and yank thinking (Jack Welch GE). This might have worked in large production and sales organizations back then, but today’s expert organizations are a different matter. When aiming for high-performance teams, why on earth should a supervisor have to choose the least excellent individuals even from excellent teams and label them as underperformers? In different cultures, the choice happens in different ways; in the USA, the quietest are chosen, in the south, the youngest, etc. This weakens cooperation, kills psychological safety, and after that, there will no longer be top teams, on which Nokia is critically dependent.
There can be several ways forward from this: 1) The chairman of the board and the CEO have a thorough discussion, the evaluation model is softened, and the mandatory distribution is abandoned. The situation calms down, and personnel focus on business. 2) Management does not back down or communicate, news reporting on Nokia’s HR issues continues, key personnel start to leave, and the consequences begin to show in business. 3) Some kind of intermediate solution, supervisors apply the practice as they see best, meaning the practice changes, but management monitors the results and believes it controls the staff, and business suffers. It is easy for an investor to choose from these three.
Was it already implemented during Simo “The Bloody’s” time? Nokia’s HR policy has probably never been particularly successful. Outwardly, Pekka Lundmark has seemed like the most modern leader. When one has looked closer at the long-term development of Nokia’s optical products business (based on Bell Labs research (InP, coherent DSP etc.), Elenion acquisition (SiPh, hybrid optics), Infinera acquisition and InP-PIC investments), as a result of which Nokia is opening up a thousand-dollar opportunity in AI-era optical/IP products, Lundmark’s appreciation rises even further. One can only hope that the current management doesn’t mess it up.
In this day and age, one can expect almost anything to turn back about 100 years. Back then, wealth and capital were heavily concentrated among a few, and the direction is the same today. This also applies to the observation made by @Oxymoron, a return to Taylorism and Fordism in a new way. Fordism, however, emphasized the importance of pre-selection, and Taylorism then focused on strict division of labor and efficiency in work organization. This is also a reputational damage when considering all the ethical responsibility to which Nokia is otherwise committed. Since Nokia has famously had annual co-determination negotiations (YT) ongoing, weren’t those enough to handle things? Nokia has certainly known how modern organizations and work communities based on workforce expertise operate, so why was it necessary to come out with such a declaration?
I myself find that at least contradictory. On the other hand, the media seems to be writing about the matter as told by a trusted person - and in such cases, the examination of the matter may be partly subjective. So I don’t doubt that he is telling the truth, but if there is some softening feature/characteristic to the matter, it may not be brought up here.
Then a little about this DCI growth engine, i.e., data center interconnect as told by Nokia. This does not yet cover the Intra DC situation, or more specifically, CPO components from Nokia’s perspective.
AI-demand-driven data center interconnect (DCI) is the fastest-growing part of the network, requiring new solutions for power consumption and space limitations.
Three key technologies define the future:
Coherent Modules (800G ZR/ZR+): An emerging standard that uses 3 nm DSP chips, offering 800 Gb/s up to 1,700 km with less than 30 watts of power. They enable the IPoDWDM model for hyperscalers.
Thin Transponders: A bridge for CSPs, enabling the use of pluggable modules while maintaining operational consistency (saving space and power by up to 40–50%).
Open OLS (Optical Line System): Flexible, modular line systems (ROADMs and ILAs) scale up to hundreds of terabits, meeting the growing DCI demand while optimizing installation space.
Three compelling AI stocks that offer attractive dividend yields are International Business Machines (IBM 0.19%), Cisco Systems (CSCO +0.58%), and Nokia (NOK 1.12%). Not only does this trio deliver dividend payouts, but they also give investors exposure to different areas of the AI ecosystem, providing diversification to an AI portfolio.
