Recently the CEO of Intel, Pat Gelsinger, made a series of comments on AMD and NVidia’s decision to develop ARM chips for the PC world. The exclusivity agreement between Microsoft and Qualcomm is now coming to an end so more and more companies will invest in this market segment in such a way as to have high-performance ARM processors ready for 2025, capable of focusing on energy saving and taking advantage of maximum compatibility of Windows on ARM with native applications and those brought into emulation by the x64 platform.
Despite Intel’s commitment to producing ARM SoCs for third parties (strategy IDM 2.0) at its factories, Gelsinger says he is certain that PCs based on ARM platform they will enjoy marginal market shares compared to those built on the x86 platform, destined to always have the lion’s share. Time will tell.
Intel’s mistakes according to Pat Gelsinger
Tough on the competition, Intel’s CEO still wanted to get rid of his own company. According to the number one of the Santa Clara company, there are three more Intel’s big failuresas he reveals in an interview published on YouTube:
- not having invested adequately in the sector of smartphone and, in general, mobile devices;
- having blocked the development of one of the first GPU designed to support applications artificial intelligence;
- not having focused in time on the supply of a large one foundry service (this aspect is increasingly managed by the company).
Lack of investments in the smartphone sector
It was 2008 when Intel presented the first ones CPU Atom designed for the world of smartphones. Atom processors were chips low energy consumption, mainly intended – in addition to mobile devices – also for emerging sectors such as netbooks. These CPUs did not achieve the desired success, also because – on the smartphone side – they could not compete with the performance and efficiency guaranteed by the SoC ARMwhich already dominated the market at that time.
Intel was too late to enter the smartphone market. When the company sought to propose its solutions, many companies had already established established partnerships with ARM chip manufacturers. Furthermore, software developers and smartphone manufacturers had now focused onecosistema ARM. These issues of compatibility and software optimization contributed to nipping Intel’s “room for manoeuvre” in the bud.
The Santa Clara company set aside its “aims” in the smartphone segment in 2016.
The sinking of the Larrabee project
Developed in the early 2000s, Larrabee is a project developed by Intel which was initially considered very promising and which had the aim of creating a new class of processors and graphics accelerators based on the x86 architecture.
Larrabee’s basic idea was to create a highly scalable and flexible processor, a GPU advanced and high performance. The intent was to rival NVidia and AMD while presenting an innovative approach, different from that adopted by traditional GPUs.
Larrabee was supposed to combine elements of CPU and GPU into a single processor, allowing not only to handle graphics workloads, but also and above all to perform generic calculations and heavy processing in parallel. Presented in 2007, the architecture was shelved in 2010.
The underlying problem is that Larrabee was looking at both the consumer market and the world HPC (High Performance Computing). If Intel had continued with the program, it would have ended up with a set of GPUs capable of supporting the modern workloads developed by generative models and by AI in general managing to compete with NVidia.
Con Xeon PhiIntel picked up where Larrabee left off but the competitive gap accumulated over the years remained unbridgeable.
Foundry activities at its factories: Gelsinger places a lot of emphasis on it
While the decision is the subject of criticism from, for example, rival AMD (according to the company led by Lisa Su, Intel will regret this choice), Gelsinger is increasingly determined to strengthen chip manufacturing on behalf of third parties at Intel factories, taking on the role played by giants such as TSMC and Samsung.
Previously, Intel had kept i production processes most advanced for chip manufacturing. Intel’s CEO, however, describes this behavior as a serious mistake. Why? Because by opening its structures to other companies, even competitors like NVidia, Intel can intercept massive flows of money that – otherwise – would be channeled into the coffers of TSMC and Samsung, proposing itself as a concrete alternative.
The capital thus raised can be used by Intel to invest not only in production but also in valuable assets research and developmentessential for the growth of the company.
Finally, Gelsinger underlined that taking advantage of market trends is rather complex because it is not always easy to understand how demand will move and make reliable long-term forecasts. Under his leadership, however, Intel still has a clear vision and seems to be aware of the developments that will allow the company to strengthen its positions and do business.
The opening image is from Intel.