As 2025 draws to a close, we look ahead to the technology and media developments that, we think, will shape 2026. AI, of course, continues to dominate (and you can see more AI predictions here), but is not necessarily the only development on the table.
Data centres take off - literally
Often when we make these predictions, some of them seem completely out of this world. This year is no exception, as we expect to start seeing off-shoring taken to new extremes with data centres being "off-worlded" into space. As reported by WIRED and The Guardian, with AI-driven demand for compute skyrocketing, the physical infrastructure required to support it may soon, literally, do the same. The benefits are clear: the sun provides practically limitless clean, renewable energy; space is (obviously) abundant, and after all, there are no local council planning applications to contend with 400 miles above sea level. There are of course significant challenges involved in jettisoning data centres into orbit (as well as material carbon costs associated with launch), not least figuring out how cross-border data protection regulations are applied when data literally crosses borders every 90 minutes. But as launch costs come down and terrestrial energy and land costs increase, the opportunities in space become increasingly viable.
"Open the fridge doors, HAL" OR "Domestic robots: from science fiction to shopping list"
Bringing things much closer to home, we expect to see an increase in mainstream consumer adoption of autonomous, smart and multifunctional robots. After decades of unfulfilled promise, 2026 may be the year household robots move from novelty to necessity. Rather than being consumer led, as discussed in TechTalks these developments are driven by greater investment and uptake in other sectors such as manufacturing, logistics, agriculture and defence. Combined with the effective use of AI, household robots are starting to look less sci-fi and more an inevitability.
Alongside the nascent normalisation of autonomous vehicles in the US (and possibly soon in the UK), we expect 2026 to see broader uptake of robotic and autonomous devices in the domestic setting. CES 2025 showcased numerous domestic robots like OpenDroids' R2D3 (which can fold laundry, load the dishwasher and do the washing up), Realbotix's Aria (pitched as a human-like companion replicating human emotions and facial expressions), and the Roborock Saros Z70 (combining autonomous vacuum cleaning with a robotic arm for moving items that would otherwise defeat most incumbent products i.e. socks). We expect 2026 to build on this and anticipate seeing the first truly cross-functional, AI-enabled domestic helpers hitting the shelves. As the number of autonomous products in use increases, so do the risks of something going wrong. The Law Commission has commenced a review of the law of product liability to cater for AI and we expect 2026 to be the first year this is tested in the UK courts.
Through the (smart) looking-glass
The theme of AI helping to crack some of the thornier consumer tech product categories continues, as we expect 2026 to be the year that smart glasses finally make it into the mainstream. Maybe. Having been through several cycles of loudly-publicised release and quiet withdrawal from the market, the alignment of several key developments may finally mean that smart glasses are here to stay. Functional HUDs and batteries that are light enough and small enough to fit on an ordinary pair of specs are now viable and can look good. This has led to the much advertised launch of the Meta Ray-Ban Display, to be followed in 2026 by rumoured releases from Apple, Snap and Samsung. But the key to why things might be different this time is, of course, AI. Combining connected, smart (and smart-looking) wearables with the capabilities of AI can enable real-time analysis of what the wearer sees and hears, allowing live directions, translations and instructions, as well as enabling interactions with AI agents informed by immediate interactions.
Smart glasses may have the most immediately profound effects in specialised scenarios, particularly where implementing domain-specific language models (DSLMs). According to research by Microsoft, these specialised models may provide more accurate results and greater adaptability in specialised settings along with, according to Gartner, improved compliance when compared with "generalist", large language models. The application of DSLMs has so far been particularly effective in settings requiring rapid answers to research-intensive questions, such as in the legal and healthcare fields. And speaking of healthcare, they are already being used to assist the sight-impaired. However, we expect 2026 to see widespread adoption of DSLMs combined with augmented- and extended-reality tech to assist with specialist, human, manual tasks, such as engineering and physical medicine.
The trust paradox - digital provenance tools will separate leaders from laggards
Widespread AI adoption has been defined by a striking paradox: according to a global study by KPMG, two thirds of people use AI regularly, and 83% expect benefits from it. Yet fewer than half are willing to trust AI systems, and in high-risk scenarios like autonomous vehicles, according to Forrester’s Global Government, Society, And Trust Survey, 2025, trust plummets to just 14%. It's not just about consumers' trust in AI decision-making, but AI providers' use of their personal information. AI-enabled social engineering, vishing, and deepfakes are proliferating. Deloitte's 2025 Connected Consumer Survey states that only around 20% of consumers see tech providers as being “very clear” about what data they collect or how it’s used. This combines with consumer unease around AI's interaction with human creativity and the creative industries, with 63% of UK adults agreeing that AI makes them value things created by humans more, according to Mintel.
Despite the gloom, we expect the paradoxical increase in AI adoption to continue, most notably in high-risk contexts such as personal finance, healthcare and legal advice, without a corresponding increase in trust. Instead, AI is, and will continue to be, used much like Google search used to be: as a first port of call, with users mitigating risks by checking and cross-referencing outputs and sources, then finally (potentially) consulting a professional as a last resort.
Digital provenance tools which consist of tools such as bills of materials, attestation databases, and watermarking to ensure transparency and trust in systems built on AI-generated content can, according to Gartner, effectively increase user trust. For businesses deploying AI, the message is clear: users will adopt your technology while simultaneously doubting it. Success in 2026 will belong to organisations that acknowledge this paradox, and build systems that are both powerful and verifiable, providing transparency tools that enable users to 'trust but verify', and accepting that AI on its own will be a first port of call, not a final authority.
Bot-to-bot commerce - the new B2B?
Beneath the consumer-facing transformation of agentic AI lies an even more consequential shift in B2B commerce. By 2028, Gartner predicts that 90% of B2B buying will be AI agent-intermediated, pushing over $15 trillion of B2B spend through AI agent exchanges, and by 2030, 20% of monetary transactions will be programmable to include terms and conditions of use, giving AI agents true economic agency.
This isn't a distant scenario. In 2026, we'll see the first wave of autonomous procurement agents comparing suppliers, negotiating prices, and executing purchase orders across global supply chains without human intervention. An AI agent managing inventory for a manufacturing plant will communicate directly with supplier agents, negotiate delivery schedules based on production forecasts, agree pricing within pre-set parameters, and commit to binding purchase orders - all in milliseconds.
The legal implications are profound and largely unresolved. When two AI agents negotiate and execute a contract, fundamental questions of contract law arise: does the agent have capacity to enter into a binding agreement? Traditional contract law requires parties to have legal capacity and intention to create legal relations. If an AI agent operating within defined parameters commits its principal to a transaction, is that commitment enforceable? The answer likely depends on whether the agent was acting within its authority - but determining the scope of that authority when the agent is learning and adapting in real time becomes extraordinarily complex.
Enforcement presents equally thorny challenges. If a supplier's AI agent agrees to deliver 10,000 units at a specific price, but the supplier's human management later discovers the price was below cost due to a data error the agent failed to catch, can the supplier void the contract? Conversely, if a buyer's agent commits to a purchase but market conditions shift before delivery, can the buyer claim the agent exceeded its authority? Traditional agency law provides some framework, but it was developed for human agents whose decision-making processes are transparent and whose authority can be clearly delineated. The question of who bears liability when bot-to-bot transactions go wrong will likely be settled through litigation before legislation. We may see cases analogous to the "monkey selfie" copyright dispute, where courts grappled with whether a non-human entity (Naruto the macaque) could hold intellectual property rights (spoiler: she could not). Here, the question isn't ownership but accountability: when an autonomous agent causes economic harm through a contractual commitment, does liability rest with the developer who created the agent, the organisation that deployed it, or the supervisor agent that failed to intervene?
The practical implications for supply chains are immediate. Leading omnichannel and e-commerce retailers have invested heavily in their online platforms, but these could be bypassed entirely when using agents. For example, OpenAI's Operator agent can 'see' and 'interact' with a webpage through typing, scrolling, and clicking, even self-correcting if it makes a mistake. If a procurement manager instructs Operator to source components, the agent doesn't visit supplier websites - it navigates them autonomously, compares offerings, and executes transactions.
This raises a strategic question for B2B suppliers: if buyers never visit your platform, how do you differentiate? The answer may lie in agent-to-agent interfaces - APIs and structured data feeds optimised for machine consumption rather than human browsing. Suppliers will need to ensure their offerings are discoverable and attractive to AI procurement agents, which may prioritise different factors than human buyers. Price and delivery speed become even more critical when an agent can compare hundreds of suppliers in seconds, but so does machine-readable certification of quality, sustainability credentials, and compliance documentation.
The $15 trillion question isn't whether bot-to-bot commerce will arrive - it's whether our legal systems will be ready when it does.
All I want for Christmas is… whatever ChatGPT tells me
Not even Christmas is immune to the unstoppable wave of AI, which may be lightening the load of tech-savvy Christmas shoppers. Younger consumers lead the charge in adopting AI tools for holiday shopping, with 44% of millennials and 42% of Gen Z shoppers likely to use chatbots and AI for purchase inspiration, compared to just 23% of consumers overall. Among those using AI for their holiday plans, 72% intend to use it for gift ideas, while 57% will compare prices and 49% seek product recommendations. ChatGPT dominates as the platform of choice, with 93% of Gen Z AI users expecting to use it for their holiday shopping. This trend reflects a broader shift towards AI-assisted shopping, as consumers increasingly turn to AI to streamline their gift-buying decisions and find better deals during the festive season.
Jingle bell stocks
This Christmas, as reported by The Grocer, Morrisons and Focal Systems are deploying AI-powered shelf-monitoring cameras to tackle the seasonal challenge of balancing peak demand with unpredictable shopper behaviour. Across each Morrisons supermarket, 400-600 AI-powered cameras capture real-time images of every shelf throughout the day, providing instant inventory checks and identifying any gaps in stock. At Christmas, when turnover in core categories like confectionery, cooking ingredients and gifting is at its most volatile, this technology provides store teams with a live view of availability, enabling them to keep shelves full during the busiest trading days of the year. The first AI deployment of its kind at scale within the UK, Focal Systems installed more than 200,000 tiny, shelf-edge, AI-powered availability cameras into 498 Morrisons supermarkets in just eight months. The technology eliminates time-consuming manual gap-scanning, which had colleagues walking entire stores daily, and instead provides continuous, accurate monitoring. By pairing these insights with predictive tools for demand and labour planning, retailers can respond to pressure before it builds, ensuring festive cheer isn't dampened by empty shelves. Just be careful when processing sensitive 'elf' data.
Humans are still here
As we progress to embedded use of AI, robotics and autonomous vehicles in 2026, humans are still needed to cross-check and override (or find those pesky stray socks). And, of course, to give legal advice (whatever the scaremongers may say).
