Artificial intelligence has quickly become one of the defining conversations in healthcare.
From clinical documentation and diagnostic support to predictive analytics and virtual care, AI is already beginning to change how care is delivered. But an equally important question is whether our healthcare environments are ready to support what comes next.
This article is drawn from our presentation at the recent 2026 European Healthcare Design (EHD) Conference in London, UK, where we examined how AI will transform not only healthcare delivery, but also the planning, design and operation of healthcare facilities.
AI is taking off. Hospitals are still building the runway.
While the conversation around AI often centres on software, it should also centre on infrastructure. The next generation of hospitals will not simply contain AI—they will need to be designed to support it.
Healthcare has made significant progress in digitization. Many Canadian hospitals now have electronic medical records, connected medical devices and systems that continuously generate data.
Yet AI readiness is not about having vast amounts of data. It is about whether that data can support real decisions and improved workflows that support patient care.
Today, hospitals face three fundamental challenges.
First, although data is captured throughout the organization, systems still struggle to reliably share and reconcile information across departments, vendors and care settings.
Second, even when information does flow from system to system, it often fails to place the data within the clinical context to make a meaningful impact to the right person, in the right place, at the right moment..
Finally, many organizations continue to treat digital systems as separate technologies rather than part of the clinical care delivery model itself.
For clinicians, improving AI readiness has the potential to reduce cognitive burden, support safer decision-making and improve clinical outcomes.
For healthcare organizations, the impact of AI allows us to plan for Hospitals that are better able to absorb surges in demand, respond to crises and evolve as technology continues to advance. The opportunity is significant—but hospitals are still building the runway.
Understanding where AI is heading
Not every AI application places the same demands on healthcare infrastructure.
Today, language-based AI is the most mature. It excels at working with text, documents, coding and summarization, and many of these applications can operate effectively using cloud computing.
The next wave is multimodal AI, which combines text with images, audio, video, and data from sensors. These systems require more processing power, tighter integration across devices and increasingly responsive infrastructure.
Beyond that lies embodied AI—systems capable of perceiving and acting within the physical environment. These applications demand the highest levels of reliability, the lowest latency and carefully designed fail-safe systems.
As AI moves from language to rich sensory information and, ultimately, to action in the physical world, the infrastructure supporting it becomes increasingly important.
Where hospitals will feel AI first
The impact of AI will not be uniform across healthcare. Different clinical environments will adopt different capabilities based on their operational needs, clinical priorities and infrastructure requirements.
Emergency departments
Emergency departments are likely to experience some of the earliest operational benefits.
Rather than relying solely on standardized triage protocols, future AI systems may compare presenting symptoms against large clinical datasets to support triage, forecast waiting room pressures, anticipate bed demand and improve patient flow.
Preparing for that future begins with the triage area itself. Sensor-rich assessment spaces, stronger integration with emergency medical services and well-defined data governance strategies will all become increasingly important.
The takeaway is straightforward: AI readiness in emergency departments starts with the triage environment and the data pipelines surrounding it.
Surgery
Operating rooms are already among the most digitally advanced spaces in hospitals.
Today's robotic-assisted surgery enables highly precise, minimally invasive procedures. Looking ahead, AI has the potential to assist surgeons through real-time image analysis, anatomical recognition and decision support, helping improve accuracy while keeping clinicians firmly in control.
Unlike traditional robotic systems, AI-assisted surgery is designed to augment clinical decision-making rather than replace it.
Because these applications are highly sensitive to latency and reliability, they also have some of the most demanding infrastructure requirements. Low-latency networking, local processing and dependable data become essential design considerations.
Mental health
Behavioural health presents a different opportunity.
AI tools may help identify patterns associated with agitation, self-harm risk, or attempts to leave a care area unsafely (elopement) earlier than traditional monitoring methods. Combined with adaptive lighting and audio systems, these tools may support earlier intervention and more effective de-escalation.
However, we believe the greatest value will come from supporting staff—not from autonomous decision-making.
Planning considerations therefore extend beyond technology to include privacy, governance, appropriate use of cameras and microphones, and workflows that ensure information reaches caregivers when it is needed most.
Pediatrics
Many emerging pediatric applications focus on improving the patient experience.
Conversational companions, interactive projection systems and adaptive rehabilitation games may help reduce pain and anxiety, personalize education, and encourage participation in therapy by responding to a child's age, abilities and clinical needs.
These applications also introduce new planning considerations. Patient rooms may require additional digital endpoints—including tablets, speakers, projectors, cameras and microphones—to support increasingly interactive care environments.
Virtual care
AI is also extending care beyond hospital walls.
Predictive scheduling, automated triage, ambient documentation and continuous remote monitoring have the potential to reduce friction before appointments while identifying patient risks earlier between visits.
The technology itself is only part of the challenge.
Successful virtual care depends on reliable device onboarding, identity management, broadband connectivity, and seamless integration with electronic health records and clinical workflows. Sensors may be located in patients' homes, but care coordination and escalation pathways remain firmly connected to the hospital.
AI resides everywhere
One of the common misconceptions about AI is that it resides exclusively in the cloud.
In practice, where AI operates depends on the application.
Training large AI models requires enormous computing resources and will continue to occur in large data centres. Once trained, however, AI models can operate in many different environments.
Where AI "lands" is a choice driven by latency, uptime, data privacy and the criticality of care.
Some applications are well suited to cloud environments; for example, gaming consoles in pediatric treatment. Others require computing close to the point of care to deliver the speed and reliability clinical workflows demand. Still others will operate entirely within the walls of the hospital to maintain local control over performance and data―in our minds, surgery is far too critical to sit anywhere other than the hospital, in order to better control the latency and reliability of data.
Rather than relying on a single AI platform, hospitals will increasingly operate multiple AI systems working together across cloud, local and hybrid environments.
Planning hospitals for an AI-enabled future
As AI capabilities mature, the planning implications extend well beyond software.
Organizations considering on-premises AI will need to think differently about power, cooling and physical space.
Greater computing capacity increases electrical demand and backup power requirements. Heat generated by dense computing environments becomes a facility design issue rather than simply an IT issue, making technologies such as direct-to-chip liquid cooling increasingly important. Higher-density equipment may also influence structural planning and floor loading.
In other words, there is no one-size-fits-all approach to healthcare data centre design.
The right solution depends on the clinical applications being supported and the operational requirements they create.
Four principles for planning intelligent care environments
While no one can predict exactly how AI will evolve, these four planning principles can help healthcare organizations prepare for what comes next.
- Design care spaces to generate data. AI depends on timely, reliable information. The physical environment should support high-quality data capture from the outset.
- Treat digital systems as essential infrastructure. Reliable networks, sensors, computing capacity and data governance are becoming as fundamental to hospital performance as traditional building systems.
- Protect capacity for change. AI capabilities will continue to evolve. Planning flexibility into today's facilities creates opportunities to adopt tomorrow's technologies without major disruption.
- Plan for distributed intelligence. AI will not exist in a single location. Healthcare organizations should expect intelligence to operate across cloud, local and hybrid environments depending on clinical needs.
Ultimately, AI-ready infrastructure means more than installing new technology. It means creating hospitals with the digital resilience to deliver safe, effective care with AI, without AI, and during the transition between the two.
That is the opportunity before us—and one of the defining challenges for healthcare planning in the years ahead.
Authors:
