Companies in every industry are rebuilding the technology they run on. The ones that plan all the parts together are putting themselves in the strongest position for the years ahead, because the main building blocks work best when they fit together from the start. Those building blocks are cloud computing, the company’s data, the software applications staff use every day, and the screens and devices people touch. Research from Deloitte’s Center for Integrated Research explains how to design them as one connected system, so quick upgrades turn into lasting flexibility.
The research describes four big changes happening to company technology all at once. Each is moving at its own speed, and each is starting to rely on the others, so a smart decision in one area helps the rest. Taken together, they show where the clearest gains are, and those that can spot the direction early can build systems that keep getting better over time and hold their value as things change around them.
The converging cloud
Most companies now run their technology across several cloud providers at once, alongside a mix of newer setups. Some processing happens in large central data centres, known as the cloud, some happens closer to where it is needed, known as edge computing, and some companies build their own dedicated AI systems, sometimes called AI factories. The number of possible combinations is growing quickly, which makes a clear plan more valuable than ever. Companies that treat their data, storage and network connections as central design choices tend to come out ahead, because the speed and control of those parts decide how well everything else performs.
The internet feels spread out across the world, yet most of its services run on just a handful of large providers. When one of them has a problem, the effects can spread quickly, which is why a growing number of companies now share their work across several providers as standard practice, an approach known as multicloud. Big platforms are themselves teaming up to share AI work and services, and Amazon recently launched a content marketplace designed to send web traffic back to the original websites. The practical step for any company is to find where it depends on a single supplier and build in a backup, so it keeps running if that supplier has problems and has more room to negotiate.
The rise of agents
A new kind of software is changing how people find information and get things done online. AI assistants can now search, compare, buy and arrange things on a person’s behalf, so the person no longer has to work through each website by hand. When many of these assistants work together across different services with little human involvement, the result is sometimes called an “Internet of Agents.” The software that coordinates them is becoming the most important layer to get right, and companies that adopt shared technical standards early can free their engineers for higher-value work, such as redesigning how tasks get done and keeping people in charge of the big decisions.
Several companies are already setting the ground rules for how these assistants talk to each other, and backing the right ones early is worth doing. Google created an early standard called A2A for agents to work together, then handed it to the Linux Foundation, a non-profit body, to keep it open and secure for everyone. Cisco has put forward its own design, the “Internet of Cognition,” meant to let agents share context and learn from one another. For all this to work reliably the assistants also need a lasting memory and a shared way to trust each other, and a research effort at MIT called Project NANDA is developing ways for agents to confirm who they are and what they can do without depending on a single central system.
Companies now actively manage and direct their data, treating it as a live resource. That raises the value of the systems that organise and protect it, such as vector databases and retrieval pipelines, which help AI find the right information quickly, and data permissioning, which sets who and what is allowed to use it. One practical approach is forward-deployed engineering, where a technology provider’s engineers work inside a client’s own systems to build and launch solutions on the spot. Working alongside the people who own each process, setting clear targets and adding safeguards from the start, helps these AI tools move safely from small trials to full, dependable use across the business.
Into the real world
Computing is moving off the screen and into the physical spaces where people live and work. Money that once went into virtual worlds, often called the metaverse, is now flowing into lighter, everyday tools such as smart glasses, devices that understand their surroundings, an area known as spatial computing, and AI that works quietly in the background as people go about their day. Counterpoint Research reports that shipments of smart glasses jumped 139% in the second half of 2025 compared with the year before. The likely result is technology that builds on the phone and brings useful information straight into a person’s line of sight.
Tools like these only work well when the connections behind them are strong. That means fast, reliable networks, processing that happens close to the user, known as edge computing, and information that updates instantly across devices. The everyday network most people never notice, made up of newer Wi-Fi standards such as Wi-Fi 6, 6E and 7, private 5G mobile networks and local connection points, decides how quick, reliable and safe these experiences feel. As AI begins to act in the physical world through cameras, location sensors and even robots, companies that invest early in solid connections are best placed to give people a smooth and responsive experience.
Preparing for quantum
Quantum computing is often treated as a far-off idea, but it already matters for how companies keep their systems secure. Powerful quantum machines could one day break the codes that protect today’s data, so work is under way on stronger forms of protection, known as post-quantum encryption. Advances in this area, along with new ways of sending information using light, are pushing companies to rethink how they handle identity, security and the accuracy of their data. Starting now lets a company build strong protection into its systems while the new standards are still being set, which keeps later costs and rework low.
Some of the tools for this job are already in use today. Shared digital record systems, known as distributed ledgers, keep a permanent and tamper-proof history of activity, which makes it easier to check, verify and audit what automated systems are doing. The World Wide Web Consortium, the body that sets common web standards, is developing ways for people and organisations to prove who they are online without relying on a single central authority. Google has said it plans to finish moving to post-quantum protection by 2029, giving other companies a clear date to plan around.
Most of the market is still early in making these changes. Only 38% of the world’s cyber security decision-makers expect to switch to post-quantum encryption within the next three to four years, according to Deloitte’s Global Future of Cyber study. The same study found that attacks powered by AI are a leading worry for security teams, which adds to the case for acting early on data protection. Progress in light-based and fibre technology is meanwhile creating testing grounds for new types of networks, so companies that move first gain a head start in learning how these will work.
Seven ages of the internet
These changes make more sense when seen as part of the internet’s longer story. Vint Cerf and Mallik Tatipamula, writing in IEEE Spectrum, describe the internet growing through seven stages, gradually moving from something people browse to something that takes action for them. The earlier stages, stretching back through the last century, include the basic internet, the mobile internet on our phones, and the Internet of Things, where everyday objects connect online.
The stage we are in now, dated to 2026, is the Internet of Agents, which matches the rise of AI assistants described above. Looking ahead to 2030 and beyond, the same map describes an Internet of Senses that uses touch feedback and direct brain-computer links, a quantum internet built for very high security, and a fully connected internet reaching almost everywhere. The value of this picture is the way it helps a company time its spending, matching today’s investments to the stage the market is actually entering.
Built to change
Good design and engineering decisions made at the start are what allow technology to keep changing later. The four changes covered here touch every part of a company’s technology, from computing power and the way AI and data work together, through to the devices people use and the security that protects everything. Because these parts are linked, the best results come from decisions that take all of them into account at once. The choices a company makes today about technical standards, devices, where it stores its data and how it builds trust will shape its technology for years to come.
A few simple questions help turn all of this into clear decisions. They include whether to rely on a few large platforms or spread the load for greater resilience, which devices and screens matter most for reaching other businesses and everyday customers, where to keep and protect company data as the wider system keeps changing, and how to avoid leaning too heavily on any single supplier.