AMD has been working on custom chips for years, and I believe that by 2026, this will pose a huge problem for AVGO and ARM.

The biggest advantage of #ASICs from the beginning was that they could be designed entirely for specific purposes; this allowed companies to focus on specific workloads, achieving much higher energy efficiency and significant cost advantages.

However, this power comes at a price: ASIC development processes are extremely long, complex, and expensive, making it impossible for every company to undertake them. Furthermore, once an ASIC is designed from scratch, it constantly needs to be improved, and each generational leap means a lot of time and a lot of cost. Although there are thousands of companies operating in the field of artificial intelligence today, the number of those with the technical capacity and financial power to design and manufacture their own chips is quite limited; those who can participate in this league are generally limited to giants like #Google, #Amazon, #Meta, and #Microsoft.

On the other hand, ASICs also have other significant drawbacks; These chips are inherently inflexible; instead of being general-purpose structures that can adapt to any task, they are designed to focus on a single problem. This makes them extremely fast and energy-efficient in a specific area, but also fragile. Because as soon as the algorithm, model architecture, etc., in the area they focus on changes, the ASICs become irrelevant.

In other words, an initial use case is chosen in ASIC design; for example, when Meta develops a chip to run, say, the Instagram algorithm on its platform, that chip can do that job incredibly fast and with low energy consumption, but it's not possible to tell the same chip, "Now go process WhatsApp data and train a large language model with that data."

AMD, on the other hand, provides both flexibility and continuous updateability thanks to its GPU-based approach; this distinguishes it from solutions optimized for only a specific workload and takes it to a wider range of use. Furthermore, AMD has a much faster technology development and generation cycle compared to ASICs, widening the gap in the long run, and I think that gap has already widened considerably.

The same dynamic applies to CPUs as well. The ARM ecosystem offers a structure that allows large-scale companies to design CPUs specifically for their needs. The core value proposition is lower cost and high energy efficiency; that is, the system is optimized to operate at maximum efficiency in the specific workload for which it was designed.

However, this model comes at a price very similar to the ASIC model: the initial CPU design process can take years and requires significant engineering costs. Furthermore, performance improvements or architectural enhancements in each new generation mean long design cycles and high R&D expenditures. Therefore, only large companies like Google and Amazon can afford this.

I believe AMD has reached a significant level of maturity in the custom CPU sector, particularly with series like the EPYC Venice and Venice-X. This allows them to combine the efficiency and custom workload optimization offered by the ARM ecosystem with the high flexibility and raw processing power offered by the x86 architecture under one roof.

In conclusion, with EPYC Venice, Venice-X, and MI455X-based Helios rack systems, they are becoming a serious competitor to both AVGO and ARM. This doesn't mean ARM or #AVGO will disappear from the market, of course; however, the main players will remain #AMD and #Nvidia. Therefore, I find it more reasonable to invest in x86 instead of ARM, and in GPUs instead of #ASICs.

The information, comments and recommendations contained herein are not within the scope of investment consultancy. Investment consultancy services are provided within the framework of the investment consultancy agreement to be signed between brokerage firms, portfolio management companies, banks that do not accept deposits and customers. The comments in this article are only my personal comments and these comments may not be appropriate for your financial situation and risk return. For this reason, investments should not be made based on the information and comments in my articles.