The Data Dash #1: Intel - When good ones go

(–) En dashes

Let’s start with Special Report: Inside Intel, CEO Pat Gelsinger - This insightful article details Pat Gelsinger’s struggles to revitalize Intel. The company faces production delays, fierce competition, and internal challenges, while increasingly relying on government support through the CHIPS Act. All this has escalated this week as CEO Pat Gelsinger retired from the company.

The US elections are over, and it seems Twitter’s influence is waning—as well within the open source community. A proposal has emerged about whether CNCF should migrate to Bluesky. The thread was locked after people began arguing, not on Twitter but on GitHub—perhaps the platform isn’t the problem, but rather the people using it? In any case, I’ve created my Bluesky account. You can find me here. Let’s see.

For the first time ever researchers crack RSA and AES data encryption in China. This has been discussed for the last few years and we are starting to see first applied quantum systems to undermine modern encryption - “This is the first time that a real quantum computer has posed a substantial threat to multiple full-scale SPN structured algorithms in use today,”

According to latest Cisco report it appears that Cisco to be emerging from its recent challenges. While product orders grew 20% year-over-year, revenue declined 5.6% to $13.8 billion. The core networking business dropped 23% to $6.75 billion, primarily due to last year’s backlog. Acquisition-driven R&D expenses increased by 19.5%. The Splunk acquisition contributed to improved margins. Data center switching showed strong performance with three consecutive quarters of double-digit order growth. Hyperscaler and service provider segments expanded by 22% year-over-year, with the top six hyperscalers each growing over 100%. This pushed orders above $300 million, positioning Cisco to achieve $1 billion in AI orders. The Splunk integration is boosting both security growth and overall margins. Though Cisco entered the AI market later than competitors, it’s now gaining momentum through Silicon One and hyperscaler partnerships, effectively competing against rivals like Arista.

Speaking of financial reports. NVIDIA has did it again. Revenue surged 94% year-over-year to USD $35.1 billion, with the Data Center segment accounting for $30.8 billion. The company saw tremendous growth across both GPU and networking segments. Net income jumped 109% YoY to $19.3 billion. In this AI gold rush, NVIDIA has been selling the picks and shovels—with just three customers accounting for 38% of sales.

This summer, a federal judge declared that Google had maintained an illegal monopoly over internet search. Now decision has been made. This case evokes strong déjà vu from the end of 1990s, when the DOJ attempted to break up another tech giant accused of monopolistic behavior—Microsoft.

Like Google today, Microsoft was found guilty of monopolistic practices, primarily by using its dominant Windows operating system to pressure PC manufacturers into favoring Microsoft applications, particularly Internet Explorer.

The Google case represents an effort to prevent the company that controls internet search from dominating whatever comes next, whether that’s AI or something else entirely. This is going to be interesting how it will playout especially now with Trump coming to the power (like in 90s where Clinton goverment was transitioning to Bush and Microsoft split actually never happened). Next one should be Adsense.

Masayoshi Son seems to have moved past his Nvidia investment and, following ARM, is now setting his sights on AI in typical Masa Son style. He is proposing to purchase $1.5 billion worth of stocks from OpenAI employees, which, according to my calculations, would give him approximately 1% ownership of the company. This could secure him a seat at the table. Based on available information, this would make him the second-largest shareholder after Microsoft.

(—) Em dashes

Are they really coming for us? Start with reading this report from FT where they are connecting latest development in robotics and AI. And now connect this report with what is BMW currently testing and what is happening in VW and will see more and more in the future as automakers will need to become more competitive between each other and mainly with China automakers.

I don’t know what is your experience but to me it feels like since COVID it is extremely hard to organise any live events. Something has changed and it feels like people are much more comfortable staying at home and not leaving their houses. One of the indicators could be that London clubs are closing at an increased rate after the COVID-19 pandemic and this newsletter sort of trying to go little bit deeper. It is funny to read and makes me wonder what implications this will have on society and marketing in general.

Intel - When good ones go…

Even though it may seem not too important, I think it could be a major catalyst for the whole industry - Pat Gelsinger has resigned as Intel CEO. Pat was a well-respected leader in Silicon Valley with a long history, not just at Intel (he joined Intel in the early ’80s) but also as VMware CEO, where he built VMware into a dominant player. As far as I know, nothing has been published about the reasons, but the guess would be that he was asked to resign by the board. He has made too many enemies. I think it is a shame because if there was someone who could save Intel, it was Pat due to his strong relationship with Intel and his leadership and engineering skills.

Intel has been dominant for the last 20 years in the semiconductor and CPU world. It still has a major market share in servers and computers like laptops or PCs. The majority of Cisco servers we have sold in the last eight years are running on Intel CPUs. Many Cisco devices like routers and switches uses Intel chips. But this is about to change….

Intel is the last semiconductor company in the world that designs products (CPUs, DPUs, GPUs) and owns multiple fabs where they make their chips and produce chips for a few other clients. This has proved to be an unsustainable business model, which may have worked in the 1990s but not in the 2010s or 2020s. Intel thought they could do everything, but AMD and Nvidia knew they could only focus on one thing - designing superb products. In the late 1990s, TSMC set the strategy to become a global leader in semiconductor fabrication - and they did. Intel is now squeezed from multiple directions by AMD and Nvidia, which are dominant with their products, and by TSMC fabs, which are 3 or 4 generations ahead of Intel fabs. In 2024, all of this broke Intel’s neck.

How did Intel get there? In the 2010s, Intel was under leadership that didn’t understand technology, didn’t have a technical background, and didn’t understand the market, so they missed two (maybe three) important things: 1) They didn’t invest enough in their fabs, 2) they completely missed AI, and 3) Intel declined to make chips for Apple iPhone (apparently due to low margins). From there, it was a fast way down from the top, and COVID, due to chip demand, exposed this even more.

Since Pat became CEO in 2021, he thought he would invest in fabs and play catch up with AMD and Nvidia on AI, but that appeared especially in the last year to be an almost impossible task. The last try just a few months ago was that Intel created two different business units - product design and fabs to keep their finances separate, but that only exposed how bad it actually is without much of a solution.

Intel is currently bleeding multiple billion dollars a quarter because of their fabs, and at the same time, their competitors are consistently taking their market share in the server and computer space. Intel CPU design is at least two years behind their competitors. Intel doesn’t have any decent products when it comes to AI (their GPU products are designed based on a design five years old), and Intel’s fabs are approximately 4-5 years behind TSMC fabs. The US government, via the Chip Act, is pumping multiple billions of dollars of funding into Intel fabs, but this could change after Trump becomes president, which can make everything even worse. Even though the US government will keep pumping money into Intel, will it be enough?

The market is currently going through a massive transition from CPU to GPU, and at the same time, we are seeing a transition in the CPU world from x86 (Intel)/x64 (AMD) to ARM (AWS/Google/Apple/Cisco Silicon One). The writing was on the wall for the last few years that this is happening, but Intel didn’t react fast enough, and maybe we are witnessing the decline of one of the greatest tech companies of the last 30 years.

Two interim co-CEOs have been announced: David Zinsner and Michelle Johnston Holthaus. Dave was Intel CFO, and Michelle was Intel CRO. This looks like they will do damage control - now I will only speculate - and I think they will be preparing Intel fab business to be completely extracted from Intel Corp and move under a new company and probably sell the product side of the business to some of their competitors i.e. Broadcom. I don’t believe their new leadership has time to wait for a new CEO or have skills to make any technical decisions - this will be about pragmatic financial decisions to at least save some money for Intel’s shareholders.