On-Device Generative AI: Unlocking True Smartphone and PC Value

The markets for smartphones and personal computers (PCs) have been in flux for over ten years. Because these devices lack "must-have" features that boost sales and provide end users an excellent value proposition, their refresh cycles have stagnated or even decreased. My organisation, ABI Research, feels that innovations in artificial intelligence (AI) hold the key to the answer, since they provide better experiences and hitherto unheard-of resources for boosting output.

On-Device Generative AI: Unlocking True Smartphone and PC Value
On-Device Generative AI 

This breakthrough takes the shape of generative artificial intelligence (AI), which was previously limited to cloud settings with enormous amounts of memory and processing power. But with to developments in hardware and software, these workloads can now be executed on PCs and smartphones, solving issues with response latency, networking costs, and privacy.

Our analysts have seen that chipset suppliers, device manufacturers, and independent software vendors (ISVs) are coalescing to enable on-device—and subsequently hybrid—AI capabilities in an effort to spread AI more broadly across personal and business devices. ABI Research projects that by 2030, there will be over 1.8 billion yearly shipments of devices that can run generative AI on them, a significant increase from the over 200,000 shipments in 2024 as generative AI becomes more and more prevalent in gadgets.

New productivity-based apps that encourage consumers and businesses to replace devices are made possible by recent advances in PC chipset technology and a range of software packages.

AI-Based Productivity Apps Boost Device Sales

According to current survey data, only a small proportion of people replace their cellphones every year or two, which might negatively impact hardware sales. The situation is much worse for PCs and laptops, where replacement rates are now more than four to five years. Up until now, the majority of AI software developers have concentrated on creating "experiences," including voice assistants that aren't very good and photo editing.

This won't be a strong enough argument for most individuals or organisations to support gadget updates. Rather, by employing generative AI within the device, "killer" productivity applications have the ability to shorten the time between gadget renewal cycles to under two years, allowing purchasers to rationalise the ROI. Vendors must provide a strong case for the applications' return on investment by citing time or money savings, for example.

How On-Device AI Transforms Consumer And Enterprise Applications

ROI-driven advantages of on-device AI include higher productivity along with enhanced security, lower latency, and cheaper networking. These advantages encourage the use of next-generation productivity apps in the consumer and business markets to power on-device generative AI PCs and smartphones.


On-device generative AI apps may assist users with a variety of activities, including tax return completion, energy efficiency, scheduling holidays, and managing smart home equipment. There are real benefits associated with these productivity apps, such lower utility expenditures and more time for leisure each day. Prosumers and artists may also be more inventive with their marketing and the creation of their art thanks to these apps (e.g., music production, video editing, etc.).

Promoting these special advantages will draw in more customers and encourage them to change their gadgets more frequently. This might turn into a long-term consumer trend that results in increased revenue when developers refine their applications or create new AI-based productivity tools.


AI programmes built on precisely calibrated models can customise taking notes, automate the generation of contracts, arrange meetings across time zones, generate meeting summaries and results, compose email drafts based on context, and more. Despite the seemingly little nature of some of these company chores, the cumulative time savings add up to a greater cost reduction than you may first believe.

For instance, a worker earning $20 per hour and receiving 30 minutes back each day would result in $2,210 in organisational savings over the course of a 231-day workweek. By implementing mobile devices with inbuilt AI capability, you may potentially achieve considerable revenue by multiplying this figure by the number of employees utilising productivity apps based on artificial intelligence.

In addition to automating tasks and saving time, on-device  artificial intelligence (AI) has the potential to open up "killer" IoT and XR applications in industries including manufacturing, logistics, and healthcare. For instance, spotty network connectivity is eliminated by the great stability of on-device AI, which is revolutionary for IoT and XR users in hardy industrial situations (such as mining, shipping, etc.). Similar to the consumer market, companies will be more likely to upgrade work devices and incorporate AI into operations as a result of the benefits of on-device AI.

Provide Software Tools For AI App Development

On-device translation Hardware alone cannot translate AI hardware capabilities into commercial success. Software innovation is essential, and to create productivity-focused apps that are optimised and "killer," developers need to have strong support. In light of this, hardware makers need to take into account the following four important software factors:

• Machine Learning Operations (MLOps) Tools: Leaders in the market need to put more money and effort into refining and optimising AI-optimized approaches including as compression, model imitation, pruning, and distillation. Optimised models make it easier to employ more complex, performant models on-device since they are smaller, consume less memory, and are more power-efficient. The secret to creating compact, contextualised AI models with leading hardware efficiency will be working with pioneers in optimisation and utilising technologies like retrieval augmented generation (RAG).

• Open Source: Open-source models provide simple access to forward-looking technology and datasets, which is beneficial for ISVs and startups. Vendors may create productivity apps specifically designed for on-device AI using these ML technologies.

• Consolidated Software Stacks: A unified software stack and an integrated collection of tools, frameworks, libraries, and technologies are necessary for efficient app development. AI applications may operate efficiently on a variety of popular architectures, including application-specific integrated circuits (ASICs), central processing units (CPUs), graphics processing units (GPUs), and other devices because to the interoperability that these software stacks ensure.

• Software Development Kits (SDKs): To minimise the work and time required for new hardware designs to be optimised, chip suppliers should promote the consistent release of new chip architectures through their SDKs. Apps will be more energy-efficient as a result of this.

Efficiency The development of AI apps will have a significant impact on future mobile device demand. Improved efficiency with a demonstrated return on investment outweighs brand-new "experiences," regardless of how awesome they are. While my team believes that productivity applications will mostly be driven by on-device AI capabilities, hybrid AI architecture—which links cloud, edge, and on-device—will emerge as the dominant paradigm going forward. Using cloud, edge, and on-device computing in a hybrid architecture is balanced according to what makes the most technical and financial sense for the particular use case.

Post a Comment

Previous Post Next Post