Ep 4: Dec 24 – Crisis or Crypto Christmas?

Starting with the chipset market, this remains a hot topic to continue to update, also this month.

• NVIDIA is keeping the rally of developing further, even if it encountered some problems with the heating of the Blackwall chipset, and even if different analysts raised the fair stock value, which is anyway still exchanged much higher than the recommended value. 
• I feel this will still keep rising. Many semiconductors, depending on the automotive market, are suffering, and this is not going to change in the short term. More in the EU Crisis sub-section.
• TSMC, as you remember, I mentioned on former newsletters how much is relevant the semiconductor supply chain is. Just quickly replied after the Trump tariffs announcement that they would not see any impact on the final product pricing. They are at this stage accelerating on their main plans for having full production (in Taiwan) in 2025on 2nm and in 2026 on 1.6nm. I’m curious, based on the big connection and dependency on the US, what will happen if they receive a push to start to produce 2nm in the US? That is something I would expect to come, especially if the friction between China and Taiwan increases. As I referred to in the former newsletters, the dependency on many players from TSMC is high, and it will be interesting to see how the market will adjust to reduce the risk. Their stock is suffering at this time, but they are, from my perspective, a good deal on the long term, having such a diversified ecosystem of customers and so advanced technology.
• On the CPU market, Intel and AMD seem to be competing, and AMD is surpassing Intel when we speak about datacenter CPU, and I still believe some consolidation will happen in the future. 
• Just reminding that Qualcomm stock is still under pressure until around Christmas (60-day deadline from ARM from the 23rd of October). You can find more in the former newsletter release. If the conclusion is successful, like I believe, Qualcomm should ramp up decent well.

Speaking about big techs also linked to AI:

• In October, we had the earnings update for the Q3 for Microsoft, Google, and Amazon, as I reported in the former Newsletter, but in November, we saw the effect of the election on the Tech companies. Amazon got a good burst, as well as initially Google, which only recently got impacted by the Chrome selling proposal coming from the US Department of Defense. Microsoft, from its side, struggled instead to gain stock traction. The market is not recognizing the ROI from the AI investments.
• Gartner, during its annual Symposium in Barcelona main event, communicated the fact that AI is tending to the end of the hype phase, which is then followed by the usual disillusionment phase. In short, this means also that within the next year, many companies will realize their first tangible results from their first AI initiatives and will have a better view of their ROI. In the section below for AI Risks, I detailed this more.
• My Thoughts: I speculate here on the fact that the former and future US president was also in a tight fight with tech companies in the last mandate around immigration bans (here, for example, an article from 2019). I feel the actual trend seems to try to take away some of the power those tech enterprises gained so far, and be sure they would not be a problem with the immigration plans. It looks to me like starting back where the conversation was left last time.
• Future US president, near the end of November, released also on his comms platform that he would once in power, apply a 25% tariff on the import from Mexico and Canada and raise the one with China for another 10%. Europe should stay, based on former exchanges on 20% tariff. December will be the time when the main US importers are going to try to get as many goods as possible before the new tariffs come into place (technically after the 20th of January 2025).

Also, news in the crypto market

• Finally, for this summary, the many messages about future US investment in Crypto made a big bounce for many crypto with Bitcoin reaching near 100K$ and many others raising their value by 2-3 times in a couple of weeks. I keep a detailed analysis of Crypto in a separate section below.

The recent US elections will put more pressure on the European Union members in terms of export tariffs to export from Europe to the US, and the general deregulations will influence how the tech services from the US will come to be perceived in Europe. Take, for example, the recent regulations of the European AI Act that will be fully operative within 2 years. Tighter control from Europe could influence the decision to prioritize or reduce the investment in this direction from US enterprises. 

• We saw the case, for example, recently Apple’s decision to postpone Apple Intelligence introduction in Europe to better comply with EU regulations. Similarly, Google is to comply with the EU Digital Markets Act on its search engine. 
• On the other side, I like to remember that the European Union has also developed a strong culture around privacy, human rights, and environmental care, and the way to build new capabilities is not by following the same path and speed as some other countries. To attach one example to this, again in November, we saw Musk asking people to upload personal medical data to X to train his AI (Grok) engine and get medical recommendations, having clearly said that such data could also be shared with third parties. The tendency to feed an AI environment with data not always completely balanced in terms of privacy and sensitive content can be a fire-back once that data is used to train the engine too. As you understand, Europe starts from a different way to attack such aspects, and the consideration around privacy is strong.
• At the same time, the car market crisis in Europe extends to other connected businesses providing automotive parts (like Bosch), and the renewable energy businesses get influenced by the new oil drilling US politics coming.

In the area of energy and how this impacts environments, some interesting news

• The energy business seems to be having an important change, influencing also all those businesses depending on it. The big techs are going to be great requestors for energy in the future, and the limits of energy deliverable through the country’s power grids are pushing the topic of finding other sources, near the datacenters, approaching micro-grids with nuclear plants and new micro-nuclear reactors able to switch off in milliseconds in case of problems. Also, Gartner, in the recent Barcelona Symposium, mentioned it as one of the key initiatives. I mentioned long ago in November’s newsletter about the possible effects I would see and some reference links and questions I raised. From that time on, for example, Amazon got blocked in its intention to push for its own nuclear production for the time being. I’m curious if hydrogen power plants will have a chance to be progressively ramping up in the overall game of CO2 reduction, increasing energy demand, increasing energy, and reducing CO2 at the same time. 
• Concern I have is instead to go back to produce energy “the old way” and introduce concepts of geoengineering, but without proper research also on possible side-effects, as there is much attention from China (at COP29 still classified as a developing market) and by the US.
• In November, there was also the COP29 in Baku with several feedback about failure to get proper alignment, missed important presence, and not sure how better is going to be supported in the upcoming years.
• It’s interesting to see that part of the energy consumption we are demanding is caused by AI’s future needs, and the hype around it could go down as we progress and leave with a case of energy surplus. 
• At the same time, cryptocurrencies are also going up, and there is a huge energy consumption to keep blockchains consistent. The future is energy demanding, and also clear that it will cause a non-linear demand and not necessarily be distributed in a uniform way across the world. 
• As I mentioned in last month’s newsletter, many AI activities remain strictly geographic specific, as no real fully distributed algorithms apply to the full machine learning process to distribute the workloads across regions.

November was a big hype on crypto. Interesting effects linked to future quantum Computing. What happened after the US election seems really interesting

• The SEC responsible in the US has no alignment with the new government strategy and is gaining more confidence in the acceptance of cryptocurrencies in a less challenging way.
• The overall Bitcoin and altcoins had a huge increase. Especially some altcoins raised 100-150% in 3 weeks.
• Behind this is a big confidence from the expected bigger investment from the US Government from next year in Bitcoin and speculations raising on building even a pension fund with part in Bitcoin (here in the UK, for example), as already started in the summer in Michigan. That seems quite a risky approach from what I see, but is telling about how some realities are trying to overcome past gaps in their financing.
• Still in November, the presence of Elon Musk in the new US leading team showed a big influence on Tesla stock (rose more than 30% in days) and then in the meme crypto Dogecoin (DOGE as stock name as his new US department). I don’t comment here apart from that each one should analyze what a meme crypto is and understand which value is behind it, if any. Different topics for those cryptos that have a real purpose.
• My Thoughts: I see a Crypto rally until Christmas and January, but also quite unstable and risky, depending on how long the US Government keeps investing in Bitcoin, and how much other cryptocurrencies will be considered
• There is an explosion of new cryptocurrencies and new meme coins, making the overall investments diluted over several solutions
• Specifically, a possible Christmas Crisis because, as much as the Crypto can be an opportunity, it will be a threat, as happened to many investors who went short of Bitcoin and lost big amounts too.
• My Thoughts: The other threat to Crypto in the long run, I see, is Quantum Computing, which will require an evolution in the crypto blockchains to be quantum hacking safe. So it is relevant that those who invest in the long on crypto are also considering that such a market will be progressively influenced by the underlying blockchain’s reliability.
• My Thoughts: Other considerations and correlations are that cryptocurrency evolution will not be without a serious energy impact, and we have already seen that energy demand, micro-grids, and new energy ways are themes quite sensible where we have a serious impact together with AI. Energy consumption is going to the stars, considering the mix of higher demand coming from AI and blockchains.

On a more general news perspective:

• Gartner reported recently in Barcelona that only 48% of digital initiatives are successful. 
• When we look around AI initiatives, the rate of failure has ramped up from several analysts and Gartner to over 80%. 
• An aspect I’m sensitive to reporting from a few months is the risk of a cost explosion linked to AI adoption if not proper ROI is done, estimating also the evolution of the costs of consuming AI capabilities as the data sources and requests will grow. I was referring to that already in October and then in the November newsletter, as my concern was that it would erode the IT ops budget. As reported now by CIO.com here, referring to Gartner analysis, I see this concern is getting more concrete, especially when we look at the expected forecasted costs. As visible from the link, Gartner analyst estimates possible miscalculation up to 1000%, so 10x the original budget cost for an AI solution from CIOs. I’m not feeling that it is a wrong estimation indeed because the cost prediction for such solutions is highly influenced by the training engine, the data quality and amount, the level of queries and sources, and many variables that in a POC could be too optimistically estimated or may miss many other conditions.
• So here is my warning about the introduction of AI, with a proper TCO evaluation and to build a proper business case in the early stage, well before the phase of production, and then later continue to monitor its evolution.
• Like I wrote last month on the rise of Agentic AI, the market of Agentic AI just started, with, for example, Salesforce advancing on its Agentforce, Microsoft announcing its Copilot Autonomous just after Ignite, SAP ramping up with its Joule, and many others. So far, we discussed around LLM, but the agentic, combining the LLM and all the RPA, and introducing a level of autonomous have a much bigger impact than what we saw so far with bare GenAI.
• At the end, the concept of Autonomous Agents is not new. It was a paradigm already defined during the time of my university studies in CS in 1998 and in my master thesis of software engineering, combining my specialization in distributed systems, I was addressing the underlying base of communication based on webservice (part of what is today SOA) with the purpose to create an interoperable, distributed, loosely-coupled architecture combining business workflow modeling, able to be distributed in the way to interact and decide. 
• Today we see the legacy with monolithic and predictable environment on one side, the loosely coupled with geographic localization in the middle, with, for example, micro-services, and the fully distributed, loosely-coupled and autonomous on the other extreme side, which is getting now more interesting with the LLM part of AI. From my perspective, we are at the start of the trip
• Gartner reported, “By 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, enabling 15% of day-to-day work decisions made autonomously“. This is a relevant statement and seems to me reliable and can help to make the maths about efficiency to gain in operations, but requires a conscious way to be introduced, and I just mentioned a few examples of risk in the AI Risks section.

Looking to future trends linked to AI:

• Agentic AI is going to transform the ERP segment in terms of part of the processes to be fully autonomous in terms of evaluation and autonomous decision-making. It’s expected to apply also to processes more crucial because based on more accurate internal data
• The CRM part is based on less precise data and not always of the same quality as ERP, so the autonomous can be connected to specific subsets of processes where possible automation could be considered. So far well developed the post sales processes, mostly customer care related. I see companies speaking mostly of chat avatar as their main AI agent, but here that is for me the real starting point. I’m sure here it will be interesting to see some of the capabilities, for example, from Salesforce’s Agentforce.
• The OT part has a big area of capability. The trend in the market is to have OT more integrated with IT, and because of the different lifecycle, there is a good autonomous agent opportunity, especially linked to digital twins, for preventive maintenance, for example
• The shared functions for me have a strong opportunity with Agentic AI. Procurement processes are more autonomous, as we already mentioned, and SAP Joule and some HR processes, especially on the hiring and more administrative tasks (linked to questions from employees, like self-service for recurring administrative tasks)
• Please remember that an autonomous agent combines RPA, LLM, and also memory for being able to create a seamless experience, realizing the automation and autonomous decision on different business workflows.

On a totally different topic, it’s always important from my perspective also to look back and remember that we live in a digital world, and if we don’t design it properly, we can land in a non-functional environment. Just I say this to keep attention when we speak about implementing IT technologies, about the side effects of some critical infrastructure. 

• Specific case I want to mention is the one of Denmark, which, in November, due to a mobile outage, got completely blocked in many critical services and operations, showing a quite dependent situation on the critical infrastructure. This is making me concern how fragile our environment would be under a byzantine case, meaning a case with the purpose to generate panic and not a failure like in this case.

The actual situation in the market, especially the European one, is clearly indicating that some industries are developing, but many industries are either restructuring or holding decisions for the US’s new strategy from next year, and a common sense of uncertainty is causing a standstill in moving. I always thought that not deciding is like deciding not to do anything and bringing consequences anyway.

There are those negative on the market speaking about a crisis like 2008, with the difference that at that time the energy cost was not as high like is today. I tend to think that a crisis is an opportunity to make things differently and take advantage of a special situation. Steve Jobs once interviewed on how to manage during crisis time was referring to the last thing he was focusing on was lay-offs (due to the effort taken to hire good people), rather than funding and investing in R&D to come out of the crisis stronger and ahead of competitors. Now, 2009 was the year in which we had a big explosion of first cloud providers and SaaS solutions started to gain wider market. So here are some reminders of why Digital Transformation is mostly needed during difficult times:

• A company’s administrative and non-operational costs can be highly optimized by automation.
• IT spend during the crisis should be funded in the area of RPA and maintenance to allow other departments to develop their capabilities of automation and guarantee supported environments during the next ramp-up.
• Business operations can benefit from automation (RPA) and from analytics consolidation to better predict future spending based on historical data.
• Business operations can improve from proper digitalization of processes and consequent effective process mining.
• Support (Shared) functions are crucial to a digital transformation because, for example, HR, Finance, and Procurement key processes need to be embedded with Ops and Sales to make efficient end-to-end processes.
• Building end-to-end business processes integrated and with data aggregated to allow consistent analysis and forecasting is crucial to achieving informed decisions.
• Crisis time and low volumes are the ideal time to close gaps in technology and reduce quickly IT risks that can be much harder to solve during peak time.
• Autonomous agents can accelerate the RPA, introducing also autonomous decision capabilities during Digital Transformation. More in the Agentic AI section.

I can’t speak about AI capabilities and Autonomous Agents without giving a note on how much is relevant to make such type of initiatives properly implemented, always tracking the risks during the full lifecycle of projects and then of operations.

At the recent Ignite, Microsoft anticipated the risks of the new Copilot Autonomous indeed coming with a set of best practices on how to avoid oversharing and risks, for example, that sensitive HR information is shared with all employees.

Here I want to remind you that:

• new technologies, especially AI-related and GenAI, can create the illusion that a thinking brain is behind evaluating what makes sense and what not to do or to share, but there is a human agency needed to guarantee proper control and progressive training of such technologies
• Any IT technology implemented without the proper attention and without preparing for “what to do in case of” is going to be risky.
• Microsoft’s recent documentation is showing most probably how much risk is perceived by the IT giant in terms of wrong data sharing that could come from enterprises once using such type of capabilities. Here is a good summary, too.
• AI solutions remain machine learning platforms, capable of predicting behavior based on how well we fed them in training, and can bring bias based on how the data are generated and fed, how some inferences are deducted, and which models are used. In my former newsletter, I tackled some of these cases.