AI Accelerating Competition Between x86 and ARM Ecosystems

Intel has released its first mobile processors based on the Meteor Lake platform (“mobile” being referenced by Intel as the notebooks/laptop sector), namely the Core Ultra H and the Core Ultra U series, which are the first of its chips to feature a Neural Processing Unit (NPU) designed to support the Graphic Processing Unit (GPU) and the Central Processing Unit (CPU) in handling Artificial Intelligence (AI) workloads. The new design is based on a chiplet architecture with multiple tiles, each accommodating a specific processing unit (e.g., GPU, CPU, NPU). The idea is to enable heterogeneous computing capable of handling a variety of AI workloads efficiently. Trading on performance improvement and reduction in power consumption when compared to competing processors from the likes of AMD, Qualcomm, and Apple, Intel’s integration of NPU into its new chips will also make it more efficient to run AI models and be capable of performing AI-powered tasks. Some laptops have already adopted the new Intel Core Ultra chip with others from major Original Equipment Manufacturers (OEMs) expected to follow suit in 2024.

While embedded AI has long found its way into many consumer and industrial sectors, one notable absence has been its use in personal computers (PCs). However, this is about to change in 2024. Where PCs have previously relied on the cloud for AI, they are being readied for it to be embedded, providing a huge improvement in data security and privacy among other benefits, especially as there are now a number of chipsets available that will help run generative AI and enable more AI-based applications. Leading the charge is Intel with its announced Core Ultra processors, but it will have to combat a surge in competition from other AI chip vendors in the PC space, namely AMD’s Ryzen 8040 chipsets, and those based on Advanced RISC Machine (ARM) architecture, specifically Apple’s M-series and Qualcomm’s Snapdragon X Elite chips.

Notably, the availability of ARM-based silicon for PCs has been available for a few years now. Since 2020, Apple has shifted to ARM-based silicon for its Macs through use of its M series of processors thereby bringing all processor chips for its mobile devices in-house, including iPhones and iPads. Apple’s switch from using Intel chips has brought both efficiency and better battery life without sacrificing performance, while also trading on its vertical ecosystem and common architecture to provide deep integration between software and hardware. Similarly, Qualcomm has targeted the PC market since 2018 with its Snapdragon 835, Snapdragon 850, and Snapdragon 8 and 7 series compute platforms. More recently, boosted by its 2021 acquisition of NUVIA, Qualcomm launched Snapdragon X Elite, a new family of chips specifically designed for portable PCs and notebooks. Built for AI, the Snapdragon X Elite platform features the custom integrated Qualcomm Oryon CPU and is capable of running generative AI models with over 13 billion parameters on-device. PCs powered by Snapdragon X Elite are expected from leading OEMs starting mid-2024.

With this growing roster of ARM-based AI chipsets destined for PCs now available from Apple and Qualcomm, the scene has been set for these challengers to upset the status quo and start to capture market share from Intel’s notebook dominance. To analyze where the strength of competition lies in the PC market, it is important to segment between the high end sector, which accounts for around 25% volume share, and the remaining low/mid-range sector with 75%. While Apple is firmly entrenched in the high-end PC market and has carved out a significant share based on its own operating system, the remainder currently uses Intel x86 on Microsoft Windows (excluding Chromebooks). It is in this low/mid-range PC sector that Qualcomm is expected to compete initially, but here it must tackle many challenges to compete effectively against Intel and gain share.

Moreover, Microsoft is also looking to launch Windows 12 with AI in 2024. As AI becomes native to Windows Operating Systems (OS), productivity tools and applications will be available that utilize AI, which will provide a massive boost to the ecosystem by enabling developers to easily create AI apps that are more productivity orientated. Microsoft has already stopped updating Windows 10 and the company will continue to support enabled devices until 2025, forcing a move to this new AI environment which will accelerate PC replacement rates.

The latest push towards embedded AI chips in the PC market is set to spark a noticeable change in attitudes toward compute in the sector, which is expected to be manifest into 2024. Led by Intel, it is the availability and choice of chipsets driving embedded AI in PCs that will help grow the market extensively. However, despite the availability of ARM-based silicon for PCs and Apple being established firmly in the high end, it is not set to displace Intel’s x86 architecture as the dominant player any time soon. For Qualcomm, particularly, it is not going to be an easy market to break as there are many challenges it must face when compared with Intel’s current offerings if it is to aim to gain some level of parity and capture market share. These market challenges include:

• Improving chip performance is essential. Despite claims to the contrary and recent chipset-level benchmarking, Qualcomm will struggle against Intel when compared with the latest Core Ultra chips for certain applications, notably system and productivity applications.
• Microsoft Windows OS has been particularly designed to run on Intel’s chips for years and the two companies have dedicated teams to maximize the system performance over Intel Chips. As a challenger, Qualcomm will have to deploy extra efforts to make its chips achieve the same level of system integration and optimization over Microsoft Windows OS.
• Confronting the legacy of x86 is another aspect for competitors to consider. This is an ecosystem that has been maturing for years, including outreach and partnerships with key PC OEMs, and is culturally established so its influence should not be underestimated.
• Comparatively weak channels to market means getting enterprises to adapt to ARM-based PCs will be a tough sell to IT departments.

Again, Windows has deep optimization with x86 and so it will take a major swing in mindset to make the switch to ARM not withstanding its limited ecosystem creation in the PC market. At best, it may take few iterations of Qualcomm’s solution to even start competing effectively with Intel at the system level integration and reach a similar level of optimization, but by this stage the market will have moved on and grown. So, even if Qualcomm chips prove to outperform Intel’s counterparts on chip versus chip level, it will be 2026 at the earliest before Qualcomm will be able to seriously challenge Intel’s market share in the PC sector. Having said this, Qualcomm may well end up securing 10-15% of this market in the mid-term, with the mid-range and low-end segments as its main targets.