How Cloud AI and Edge AI could Transform Printing

June 6, 2024
Article, Artificial Intelligence, Artificial Intelligence, Cloud, Trends

Quocirca’s AI study reveals that 66% of organisations are actively using AI in their organisation, rising to 70% in large enterprises and 76% in the US. Generative AI models are the top areas of investment, and overall, an average of 13.6% of IT budgets are being devoted to AI (rising to 15% in the finance and professional services sectors), with 84% expecting this to rise further next year.

However, security and privacy concerns abound, often because AI relies on cloud infrastructure. Is this set to change with the emergence of Edge AI (think AI-powered PCs that run LLMs locally through NPUs such as Microsoft Copilot + PCs), and will this signal a return to on-premise environments to capitalise on secure AI platforms?

Hybrid approaches to the cloud

According to Quocirca’s Cloud 2024 study, a hybrid approach to the cloud (some cloud usage alongside some on-premise infrastructure) will remain prevalent over the coming years. While just 3% report that they are fully in the cloud today, this is set to increase to 18% of organisations by 2026. However, the trend is hybrid – with 80% operating a hybrid environment now, dropping only slightly to 70% by 2026. This reflects the complexity around cloud deployments, often used in a piecemeal approach. For instance, the SaaS solutions organisations use may be hosted on a hyperscaler, such as AWS, Azure, or GCP, but can also be on a managed service provider’s (MSP’s) own cloud. Integrating across different platforms is not always easy – and less so when an organisation is trying to move away from any on-premise hardware to a full-cloud solution.

Meanwhile, concerns around security and availability mean on-premise environments may still be favoured for business-critical or highly regulated industries.

Cloud as an enabler for AI

The cloud and AI are inextricably linked, with AI investments boosting spend in the cloud. This is reflected by continued growth among the main hyperscalers. AWS reported that during January to March 2024, revenue grew by 17% and reached a $100 billion annual run rate for the first time. Microsoft Azure and Google Cloud grew by 31% and 28%, respectively, with Microsoft reporting that AI services had seven percentage points in growth to Azure, up from six percentage points in the Oct-Dec quarter.

The cloud is making AI accessible to all, with many providers offering AI services that are built in. Cloud-native technologies such as OpenAI, Azure Open AI, GPT, and DALL·E rely on highly compute-intensive resources and storage, which the cloud readily provides. According to Quocirca’s research, Gen AI ranks as the top area for AI investment, with 45% of respondents identifying it as their primary focus (rising to 50% in the US).

In addition, the cloud provides the vast amount of data and processing power that AI requires for analysis to create suitable outcomes, and AI can also be used to help optimise cloud usage and automate tasks such as resource allocation and cost optimisation.

Security and privacy barriers to cloud-based AI

Even though cloud computing offers advantages for AI, there are some cloud-related hurdles to AI adoption. In Quocirca’s study, 31% cited security and data privacy concerns as a result of storing and processing sensitive data in the cloud. Many companies are apprehensive about putting their data, especially proprietary information, on cloud platforms.

Meanwhile, regulations around data storage and privacy might make cloud-based AI a complex issue, particularly for organisations such as the financial sector, which has stricter rules about data residency, limiting their cloud options.

Latency issues may also be a challenge, especially for real-time AI applications in which which latency in data transfer between devices and the cloud can be a challenge.

As cloud security improves, regulations adapt, and AI technology becomes more flexible, these barriers are likely to become less significant over time. However, in the meantime, the use of on-premise environments may well prevail.

The Rise of Edge AI 

Organisations that want to integrate AI into processes without sacrificing data privacy may choose a hybrid model. This enables them to leverage the strengths of each environment. For instance, computationally intensive tasks or those requiring real-time processing might stay on-prem, while data analysis or training less-sensitive models could move to the cloud. This approach to deploying AI models on their own servers is known as Edge AI.

Potential applications in the print sector include:

• Predictive maintenance. Built-in sensors can analyse data on ink/toner levels, paper jams, apnd component wear. Edge AI can predict potential issues and trigger alerts before breakdowns occur, minimising downtime and service costs.
• Security enhancements. Edge AI can analyse print jobs for suspicious content such as malware or confidential information. This can prevent unauthorised printing and data breaches.
• Real-time optimisation. Sensors can monitor printing parameters such as temperature and ink usage. Edge AI can adjust settings on the fly for optimal print quality and efficiency.
• Personalised user experience. User preferences and past printing habits can be stored locally. Edge AI can personalise printer settings, such as default paper size or duplex printing, for each user.

This approach comes with downsides – it can be more complex and costly to manage on-premise infrastructure, and computing resources will not be as scalable as a cloud infrastructure. Only with open cloud and a multitude of SaaS options can AI work to its maximum capabilities.

Combining Cloud AI and Edge AI

The optimal approach is to combine both to assist with areas such as advanced threat detection, in which Edge AI can identify potential security threats on the device, while Cloud AI can analyse them in a broader context, providing a more comprehensive security solution. In terms of continuous learning and improvement, Edge AI models can be updated with data from the cloud, while Cloud AI models can be refined based on insights from Edge devices, creating a continuous learning loop for improved printing experiences.

Guidance for print manufacturers and partners

Print manufacturers are already deploying AI to some extent ‘at the edge’. Vendors deploy AI algorithms directly on printer equipment and sensors to monitor device health, detect defects, and optimise production processes, although in many cases this relies on centralised servers. Deploying on-device AI cybersecurity on printers and MFPs will be key to addressing the fact that such Edge devices are susceptible to malware infections, cyberattacks, and remote exploitation if not properly secured. Implementing antivirus software and intrusion detection systems as already offered by print vendors today will need to be part of every company’s Edge AI strategy.

Meanwhile, the potential to use Cloud AI across the print environment is yet to be fully realised. While cloud print management solutions may use AI to help with advanced analytics on device usage and reporting, it has yet to take advantage of private or enterprise LLMs, which can help with document summarisation, printing formats, etc.

As customers evaluate their use of AI across their business, print suppliers need to ensure print solutions – whether hardware or print management – are used to address both the AI and cloud maturity of their customer. This means understanding customer needs around cost, availability, data privacy, and device management.

MPS providers will need to ensure that information security is paramount and everything is dealt with in a manner that is transparent and acceptable to users. Users will need to be given the choice to opt into such services – and opt out whenever they wish to.

Guidance for end users

The integration of AI within print devices is highly variable depending on the print vendor. End users need to determine exactly what AI features and functionality they are looking for within their print fleet and make their product selections accordingly. Having decided to incorporate AI within the printer fleet, end users must be mindful of device security and network integrity.

Deploying on-device AI cybersecurity on printers and MFPs will be key to addressing the fact that such Edge devices are susceptible to malware infections, cyberattacks, and remote exploitation if not properly secured. Such devices need to be considered an integral part of a company’s cybersecurity strategy and protected accordingly. A secure and robust network is also essential where devices rely on the network to communicate with each other and the cloud. If the network fails, the device can become compromised or inoperable.

As AI continues to evolve, selecting the right infrastructure, whether Edge AI, Cloud AI, or a hybrid approach, will be key to optimising the scalability, efficiency, and flexibility of the print infrastructure. Overall, Edge and Cloud AI have the potential to further transform MFPs and printers into intelligent end points, offering increased efficiency, security, personalisation, and cost savings.