This year, AI will be shaped by data centres, geopolitics, and the workplace

Do you need a bold speaker that delivers an actionable keynote for International Women’s Day? I’m your person.
Some of the themes of my most celebrated talks:
- How sponsorship is the missing magic bullet for career success
- Women and AI: Transforming barriers into opportunities
- Counterintuitive secrets to a thriving career in tech
- How patriarchy fosters perfectionism, self-criticism and self-doubt and what you can do about it
If you want your audience to leave energised and empowered to act, get in touch.
Last year, I published my 25 AI predictions for 2025 — and I’m happy to say many of them turned out to be strikingly accurate.
So this year, I’ve picked up my crystal ball again.
In this article, I unpack:
- The AI shifts that will really shape 2026.
- What’s hype — and what’s actually coming.
- What these changes mean for leaders, careers, and power.
- Strategic foresight you can act on now.
If you want to get ahead of the curve — not chase it — this article is for you.
“The future is already there — it’s just not evenly distributed.”
My Predictions
- Workers will be accountable for AI success: In 2025, executives were responsible for AI success. This year, workers will be. Expect organisations to roll out KPIs to boost the use of chatbots and other AI tools, and to monitor who’s using the LLM licenses they purchased.
- Expect employees to be blamed for AI underperformance: We’re already hearing Sam Altman and Satya Nadella say that if AI isn’t providing value, it’s because workers aren’t using it correctly. Look for the likes of OpenAI and Microsoft showering paid customers with free or low-cost training. And not only them, the UK government just blamed workers’ lack of confidence and enthusiasm around AI on a skill gap.
- The big AI scapegoat: companies will disguise layoffs as AI efficiency gains. Last year, we heard tech companies explain layoffs as a consequence of AI adoption—even though they couldn’t prove it. This year, that trend will extend to other sectors, which will blame “AI adoption” for layoffs truly stemming from low performance and challenging business conditions.
- LLMs as top cybersecurity risk — 57% of employees admit to concealing how they use artificial intelligence at work. That means data breaches are set to skyrocket, now accelerated by the use of AI agents by workers. Also, LLM-based agents can automate a significant portion of cyberattacks. All that in addition to what we’ve seen in 2025: LLMs make it easier to generate convincing phishing messages in different languages and to develop malicious code. And it’ll continue the constant cat-and-mouse game between AI companies trying to prevent their models from misbehaving and malicious actors devising ever more creative jailbreaking prompts.
- Workslop will skyrocket —About 40% of US workers claimed in 2025 that they were victims of Workslop, AI-generated work content that masquerades as good work, but lacks the substance to meaningfully advance a given task. The cost? About US$9M Annual cost for a 10,000-person company. 2026 will be the year workslop begins to be discussed openly and workplaces set norms to mitigate it.
- OpenAI wins with ads, not shopping assistant — Gemini is going to become the default “generative AI marketplace.” In 2025, OpenAI learned how difficult it’s to set a payment structure, whilst Google simply started scaling what it already knows how to do. This year, OpenAI will finally take a lesson from Facebook and Google’s playbook on how to make billions: ads. Be prepared to pay for ChatGPT to recommend your website.
- Chatbots (and other AI tools) take over healthcare — This year, OpenAI already released ChatGPT Health — a “dedicated health experience” that binges on your health data records— and Anthropic announced Claude for Healthcare, an overhyped new set of tools and resources that healthcare providers, payers and consumers can use for medical purposes. But it’s more than that. We’ll see huge investments from pharma/biotech companies to hospitals and governments in AI platforms and tools that promise shorter drug discovery cycles, more efficient clinical trials, and “efficiently” allocate resources — human or otherwise.Photo by Mohamed Nohassi on Unsplash.
- AI-washing: Automated ERP is the new agent — “Just Walk Out” was an Amazon technology installed in many of its physical grocery stores, promising customers could simply pick their items and leave. It was then discovered that approximately 1,000 workers in India were manually reviewing nearly three-quarters of transactions. What about agents then? We’ll see the beautification of automated ERP (Enterprise Resource Planning). Basically, our old ERP will be “wrapped” in a chatbot that interacts with the user, which then kicks off the automated, reliable, and invisible ERP system. We’ll call that “agent.”
- China and Open Source — In 2025, I anticipated it’d be the year where China — with DeepSeek at the top — would claim their rightful place in AI. I was right. This year, China will flood us with open source AI models. What better way to tank Big Tech AI revenue ambitions — and US AI supremacy aspirations — than to make AI tools free?
- Two-tier services — I’ve been working in software companies for two decades in customer-facing roles (training, professional services, customer support) and not even once has a person asked for a chatbot. On the contrary, if I had £1 for each time I got a request of “I want to talk to a person on the phone,” I’d be rich by now. Has AI changed that? Nope. When we’re in distress, nothing helps to regulate our nervous system like another human. So what’s the role of AI chatbots? Cater for those who don’t have the money to pay humans. In 2026, we’ll see the chasm widen between AI and human support: humans for the rich, AI for the rest.
- Predictive maintenance — With claims of up to a reduction of 73% in infrastructure failures, cutting maintenance costs by 10–40%, and reducing downtime by up to 50%, AI predictive maintenance — a data-driven approach to predicting machinery failure and making proactive repairs — is revolutionising industrial operations. In 2026, we’ll see a surge in the adoption of predictive maintenance and virtual twins. Why? Rising cost of unplanned downtime (e.g., global supply chain constraints, increased replacement part costs), labour shortages and skill gaps, the explosion of equipment with sensors, telematics, and digital interfaces, AI maturity (computer vision, IoT, ML models), and pressure to improve uptime and operational efficiency.
- Data centres: Location, location, location —As with housing, where you decide to build your data centre will have the greatest impact on its viability and sustainability. The proof is in the pudding. Currently, data centres in Silicon Valley are idle because Santa Clara can’t keep up with surging electricity demand. Also, those in the middle of nowhere are struggling because, when used for inference, they perform better when built near the demand. Moreover, data centres in cold ambient conditions across the Nordic countries enable nearly year-round “free cooling,” reducing the need for energy- and water-intensive mechanical cooling systems such as chillers.
- Deceleration of the number of data centres built — This year will see a slowdown in the trend of the number of data centres being built due to the exponential rise in activism. For example, over the last two years, $18 billion in data centre projects were blocked, and another $46 billion of projects were delayed amid opposition from residents and activist groups in the US. In Brazil, the Netherlands, and Chile, locals have also pushed back. The other factor contributing to the deceleration is their increasing classification as critical infrastructure, prompting more controls and monitoring.
- Data centres will continue to rely on oil and gas — Amid economic and geopolitical uncertainty, governments will deprioritise sustainable energy solutions. For those doubting the shift, remember the US military intervention in Venezuela. With AI FOMO escalating, governments and companies will continue to rely on oil & gas for “right now” as they invest in nuclear (including small modular reactors) and renewables “for the future.”
- Data centres go to space —Last year, Google released a paper proposing space-based “data centres” consisting of many solar-powered satellites networked via free-space optical inter-satellite links. They illustrated the basic approach to formation flight via an 81-satellite cluster with a diameter of 2 km. And they were not the only ones. Musk and NVIDIA have also found merit in the idea. With all the hype around investment in space right now (e.g., SpaceX’s IPO), space data centres appear to be a cunning way to attract “responsible” investment under a “sustainability” veneer.Image by gene1970 from Pixabay.
- Quantum computing for data centres — With its ability to solve complex problems beyond the capabilities of classical systems, quantum computing appears to be the answer to our thirst for more “compute.” While building quantum computing data centres is not trivial — energy, security, integration with existing systems — this year, we’ll see substantial strides in this direction.
- Cooling technology — Both classical and quantum computing data centres rely on cooling technology to avoid melting GPUs and attaining near absolute zero temperatures to function, respectively. At the same time, with climate change, cooling is becoming more difficult. As a result, innovation and investment in data centre cooling will soar.
- Big Tech becoming energy companies — Last year, xAIx formed a joint venture with a company which rents natural gas turbines to data centres and oil and gas companies. Meta agreed to purchase electricity from reactors from Sam Altman-backed Oklo and Bill Gates-backed TerraPower. Tech Bros have recognised that energy is the real superpower and will continue to extend their reach in 2026.
- Chips, China, and Taiwan — Servers with GPUs and processors that have silicon chips are at the core of data centres. TSMC, based in Taiwan, produces over 60% of the world’s semiconductors and just under 90% of its most sophisticated chips. On the other side, China has made its intentions about the island very clear. Cross-strait tensions between China and Taiwan will threaten the chips’ supply chain and drive spikes in prices.
- The rise of TPUs — GPUs have been the turf of two players: NVIDIA and, to a lesser extent, AMD. One of the key reasons NVIDIA has the upper hand, in addition to the hardware, is that it has developed the software that makes it easier for developers to use GPUs for parallel programming (CUDA), making it the “default.” At the end of last year, Google began working with Meta to make their AI chips — called TPUs (Tensor Processing Units) — run better in Pytorch, making them compatible and developer-friendly for customers who have already built their tech infrastructure using PyTorch software. This year, we’ll begin to see the fruits of that aggressive plan to compete with NVIDIA.
- Speculative bubble, AI bailouts, an innovation stimulus — There is an undeniable huge gap between valuations, revenue, and investment (e.g. data centres) for companies selling generative AI, even the FAAMG. For example, neither OpenAI nor Anthropic has a robust business model that explains how they’ll cover the revenue shortfall for 2030 for their investors. This has all the markers of the bubble. The question is, will it burst in 2026? My prediction is that it won’t. Bubbles do not suddenly explode; they are maintained by narratives before finally shattering. Whilst the mirage of “AGI will be there in 2027” has been dismantled, there are still other narratives that tech is providing to sustain the FOMO and governments are happily buying them as a “geopolitical” leverage. But how to bridge the financial gap between valuations, revenue, and spending? We won’t call them bailouts; governments will make us foot the bill, camouflaging them as innovation stimulus packages.Image by Alexa from Pixabay.
- New AGI: World models and continual learning — Sutskever, LeCun, and now Hassabis have confirmed what many of us knew all along: LLMs are not the path to AGI (even if that existed). What to do next? Reinvent our vision of AGI to sustain the speculative bubble. Two terms will become the default when promoting AGI in 2026. The first, world models. This concept is not new, but it’s more useful now that everyone understands that LLMs “don’t think” and, very importantly, they don’t have a sense of the world (e.g., gravity, anyone?). This is not an LLMs problem but a deep learning problem. The solution? World models — the concept that you can infuse AI with physics and spatial properties. As many others, NVIDIA and Google are already in the case. The other buzzword promising AGI will be “continual learning,” mentioned by Ilya Sutskever in his appearance on the Dwarkesh podcast last year. That is, a model that, like humans, can continually acquire new knowledge and keep pace with the real world’s unpredictability without forgetting prior knowledge.
- Anti AI regulation driven by tech, FOMO, and geopolitics — Big Tech has successfully lobbied its way through AI regulation across many countries — US, UK, EU, Brazil — stopping, weakening, and even rolling back digital rights. Governments have told us that data rights hinder economic growth and innovation with regard to AI, promised that AI will protect free speech and national values, and pledged to remove red tape and onerous AI regulation. Hence, in 2026, we’ll see further regulatory weakening. What will we do about the threats AI poses to children, then? Instead of regulating tech, we’ll default to increasingly regulating children’s access to it worldwide. We’ve been there for alcohol and tobacco.
- Robotics will bloom — Robots that use AI will become more ubiquitous. Those robots have greater autonomy thanks to their ability to process large datasets, detect patterns, anticipate failures, and use computer vision to identify objects. Companies such as Amazon and Tesla have been very open about their robotics progress and ambitions and now even OpenAI is investing in robotics. This year will see robotics taking a bigger role in driving operations across sectors.
- Best AI investment: Defence, espionage, law enforcement — Wars, political and social unrest, and massive increases in defence budgets will make 2026 a great year for AI companies that provide surveillance and defence. Palantir (the “OS” for governments) will keep winning lucrative deals with “intelligence” and law enforcement agencies around the world. And defence firms like Anduril, a drone and autonomous weapons startup that doubled its valuation to $30 billions last year, will flourish. The explosion of companies such as Clearview and Flock Safety that rely on AI to deliver solutions for public safety and security will continue. Even the UK has now announced plans for an AI centre for policing.
- AI for loneliness — Last year, we finally confronted the reality that a massive section of the population has been relying on AI chatbots for support — from intimacy to healthcare and career advice. The reasons are multiple. Chatbots are free or cheaper than humans, offer a sense of judgment-free exchanges, and encourage us to do what we want. The downside? Deaths by suicide, biased vocational advice, and incorrect medical treatments. The increase in the cost of living, job insecurity, and geopolitical unrest will boost our reliance on AI chatbots to give us (almost) free solutions and provide support for our problems. Given the syncophantic nature of those chatbots, we can expect further reinforcement of loneliness, entrenched beliefs, and bias.
BONUS: A wish
AI accessibility becomes the new normal — I’m stealing one of the five trends to watch in 2026 by the accessibility expert Sheri Byrne-Haber, who posits that
“In 2026, AI-assisted accessibility testing moves from optional to expected. Organizations that do not incorporate AI-enabled accessibility testing will struggle to keep pace with modern development cycles.”
I wholeheartedly hope so.
Back To You
You’ve now read my predictions. I want to hear from you
1.- Which ones do you agree on?
2.- Which ones do you dispute?
3.- Which ones did I miss?
WORK WITH ME
Do you need a bold speaker that delivers an actionable keynote for International Women’s Day? I’m your person.
Some of the themes of my most celebrated talks:
- How sponsorship is the missing magic bullet for career success
- Women and AI: Transforming barriers into opportunities
- Counterintuitive secrets to a thriving career in tech
- How patriarchy fosters perfectionism, self-criticism and self-doubt and what you can do about it
If you want your audience to leave energised and empowered to act, get in touch.






















































































































































