The upsurge in utilization of artificial intelligence (AI) in the last couple of years has significantly transformed not just software but also hardware. As AI adoption continues to progress, computer manufacturers have discovered in AI a chance to enhance end-user devices by providing AI-specific hardware and marketing them as “AI PCs.”
Pre-AI hardware, tailored for AI
A few years ago, AI often relied on hardware that was not explicitly crafted for AI purposes. For instance, graphics processors played a pivotal role. Nvidia Graphics Processing Units (GPUs) are essential in AI because they manage parallel processing efficiently, which is essential for machine learning and deep learning. Their structure facilitates simultaneous computations, rendering them more efficient than CPUs for AI model training and inference.
Another key type of hardware is the Field-Programmable Gate Array (FPGA) from Intel and other firms. An FPGA is an integrated circuit (IC) that can be reprogrammed multiple times. Such adaptability makes it perfect for AI chores. FPGAs hasten deep learning and machine learning tasks. They furnish options for hardware customization that mirrors the behavior of GPUs or ASICs.
FPGAs can be incorporated with popular AI frameworks like TensorFlow and PyTorch utilizing tools such as the Intel FPGA AI Suite and the OpenVINO toolkit.
FPGAs are utilized across various sectors like automotive, healthcare, and others. They find utility in edge computing scenarios where AI capabilities must be deployed near the data source to expedite decision-making and minimize latency.
Yet another kind is Application-Specific Integrated Circuits (ASICs). One instance is the Tensor Processing Units (TPUs) from Google. TPUs are custom ASICs devised by Google to accelerate machine learning workloads. They are optimized for TensorFlow and extensively employed in Google’s data centers.
How the groundbreaking generative AI revolution overhauled hardware
The unveiling of ChatGPT by OpenAI on November 30, 2022, reshaped the populace’s and industry’s interaction with AI. ChatGPT swiftly garnered immense popularity, captivating over one million users within five days after its launch. By January 2023, it had amassed 100 million users, propelling it as the fastest-growing consumer application ever.
Most significantly, ChatGPT’s extraordinary success in mainstream culture steered venture capital towards AI startups. Tech giants like Microsoft, Google, and Meta hastened the enhancement and public availability of their offerings, and Silicon Valley swiftly observed the emergence of firms like Anthropic and Perplexity extending AI solutions.
Presently, PC manufacturers are investing in AI-enabled PCs that stress on hybrid AI and on-device intelligence. The embedding of AI into personal computers is facilitated by the advent of specialized AI chipsets, like neural processing units (NPUs), which amplify PCs’ capability to carry out AI tasks locally.
This transition is anticipated to make a considerable impact on the PC market. As per Canalys, around 60% of PCs shipped by 2027 will be AI-capable.
Explore AI cybersecurity solutions
How are AI PCs distinctive?
AI PCs are structured to proficiently execute AI workloads by using a blend of CPUs, GPUs, and NPUs, allowing them to manage activities like generative AI models more effectively than previous PC iterations. This optimization empowers AI PCs to operate AI applications with elevated performance, energy efficiency, and data privacy by processing data locally instead of relying on cloud-based alternatives.
Some critique this class as a marketing ploy and highlight that many end users engage with generative AI via cloud-based chatbots.
Currently, the public perceives AI as large language models (LLMs) functioning in the cloud and leveraged as chatbots. As time progresses, the power and utilization of AI by end users will progressively shift towards integrated functionalities and AI-enhanced applications.
According to Gartner’s Global Chief of Research, Chris Howard, AI will also encompass more small language models (SLMs) fueling non-chatbot applications operating closer to the edge rather than the cloud.
AI processing will increasingly materialize nearer to the user and at the edge. This signifies that the trend of AI-dedicated hardware will only expand.
Microsoft AI
A standout development is the unveiling of Microsoft’s Copilot+ PCs, a novel category of Windows PCs explicitly designed for AI chores. These PCs incorporate new silicon capable of delivering over 40 trillion operations per second (TOPS), ensuring all-day battery longevity, and access to advanced AI models. The architecture of these devices integrates a high-performance NPU alongside the CPU and GPU, augmenting their AI capabilities. This setup enables fresh experiences like real-time AI image creation, language translation, and advanced search functions like the “Recall” feature, which logs and scrutinizes device activity to amplify user engagement with AI models.
Microsoft has also partnered with renowned OEM collaborators, such as Acer, ASUS, Dell, HP, Lenovo, and Samsung, to introduce these AI-enhanced devices to the market.
Apple AI
Apple has implemented diverse hardware modifications to accommodate and empower AI functionalities in its devices. A notable breakthrough is the incorporation of Apple silicon, precisely formulated to handle advanced AI processing. This involves the use of specialized neural engines in devices like the iPhone 15 Pro, which are tailored for AI responsibilities such as machine learning and natural language processing. These neural engines boost the efficiency and speed of AI operations, enabling features like real-time language translation and image recognition.
Google AI
Google has instigated several alterations in its hardware to accommodate AI. A significant stride includes developing and embedding its proprietary hardware to bolster AI models like Gemini. This denotes a shift from depending on external chips to leveraging in-house technology to enrich AI capabilities.
Google has even reshuffled its internal units to integrate AI more effectively across its products. This restructuring has engendered the formation of a new Platforms and Devices team, consolidating diverse Google products like Pixel, Android, Chrome, and ChromeOS under unified leadership. This maneuver aims to hasten AI integration and enhance the synergy between hardware and software.
The AI hardware evolution
The widespread generative AI revolution commenced in November 2022 and induced substantial hardware transformations to cater to demanding AI scenarios. The recent assimilation of AI in hardware is merely the initial stage for what lies ahead. We can anticipate AI-specific hardware permeating beyond PCs and smartphones into wearables, Internet of Things gadgets, and more.