Annual outlook

1. Technology

The year begins with steady advancement (100%). Some curves look exponential, some look linear, but much of it feels flat. Evaluating frontier models gets harder. The most interesting developments happen within long-horizon agentic tasks that take minutes to execute, hours or days for humans to reproduce, and at least minutes to verify. This makes development feel slower and slower for the general public but masks true progress in complex work (90%).

The people who did not care which model they use still do not care. The people who did care still do, but it is increasingly voodoo, and blind tests reveal that most people cannot tell the difference anyway. Telling models apart is no less important, but much harder (90%).

Sometime around summer, something remarkable happens: the world's first major AI discovery makes mainstream headlines. Perhaps a mathematics millennium problem, perhaps something bigger. It is not going to be immediately useful, but experts take note, and as a result, AI stocks see a bump (50%).

The pre-training data wall emerges (40%). Labs become tight-lipped, but we see one or more acquisitions clearly aimed at locking down archives or publishers for training data (60%).

Voice finally clicks. It becomes the primary modality for 10–15% of users. Not because it is new, but because latency and awkwardness finally disappear (60%).

Like reasoning emerged in 2024–25, a new capability shift arrives in 2026. Maybe chain-of-thought moves into continuous vector space. Maybe computer use or code generation happens in 90%+ of interactions. Maybe a lab will solve continual learning. Whatever form it takes, we get a new scaling law or architectural breakthrough that changes what is possible. But it is not going to change the previous prediction, that most people will not care – and evaluating whether this breakthrough is helpful is complex (50%).

2. Business

Bubble talk continues, but the bubble does not pop (100% – yes, you read that correctly). The underlying growth trajectories remain strong enough to carry the weight. If a correction comes, it will be in 2027 or later (70%).

The stock market still delivers drama. At a handful of legacy companies, a lack of AI progress becomes a board-level issue, prompting dramatic moves to push through change. This will involve CEO firings, restructurings, sell-offs, or similar actions. The stocks in question will take hits, and for most, it will be too little, too late. Catastrophic declines will come later, but 2026 will be the first year we see the signs (70%).

More AI unicorns arrive. Where 2025 gave us a handful of instant unicorns (Lovable, Cursor, Thinking Machines), 2026 will produce 10–15 more (70%).

Consumer AI spending keeps climbing. In 2026, power users will spend €150–200 per month through premium tiers and best-of-breed stacking. Enterprise spending doubles to around €50 per employee per month, though it will be unevenly distributed. Some professional developers will burn through €170–860 per month (and it will be worth it), while some office employees will still be on internal solutions with OPEX of less than €1 per employee per month (70%).

The first small but real labour market effects appear. Amid the noise of everything else, this remains a relatively quiet story. However, economists at central banks and ministries take note. In the halls of politics, actual plans for labour market effects start to take shape, but we will not know the full impact yet, so evaluation will have to wait (30%).

Some companies realise they can truly build their own tools now (50%). In-house applications that are genuinely useful start appearing out of nowhere. Deploying and governing this software sprawl becomes a key challenge. There will be many warnings and naysayers, but a big theme will be individual employees and small teams making small applications for themselves that are genuinely useful. The nay-saying governance people will be right in 2027, but 2026 will be a party for the little vibe-coded application (80%).

Regulation remains largely theoretical, and the industry remains, for better or worse, self-governing (80%).

3. Culture

The companion question gets personal. By the end of 2026, most people will know someone who has what can only be called a ‘relationship’ with an AI – that is, they would be genuinely sad and perhaps even depressed if it were taken away from them. It reveals something about human nature we were not ready to discuss (60%).

The mental health discourse catches up. There will be studies on both sides – AI helps loneliness! AI creates isolation! – and moral panic will begin. The nuance gets lost, but at least we will be talking about it (75%).

AI slop keeps growing, and people keep saying they hate it (this one is the easiest 100% I have to offer!). But it is wildly successful by engagement metrics (90%). The slop gets better too, good enough that it stops being an interesting topic for discussion. It is just content now (60%).

There are scares involving impersonations of public figures, but nothing prompts real action (90%). A fully AI-generated song breaks into Spotify's global top 10 (60%).

By December 2026, what once felt extraordinary about AI will have become a sort of 'new abnormal' (30%), meaning that people will grow accustomed to the constant change and often just shrug, saying, “Well, what just happened? Was it really that wild? It’s just AI, after all.” But it is important to remember that the current pace and impact of AI development are not normal. This rapid transformation has been unfolding for three years, and though we can become numb to it, this moment really is far from ordinary.

While people will talk about AI fatigue, development continues with no slowdown in sight, and the conversation goes on regardless. This is where polarisation begins, and we might see the first anti-AI demonstrations (10%).

No matter what happens in 2026, we at Implement will continue to advocate for responsible AI development and deployment, ensuring that innovation aligns with ethical principles and delivers a net positive impact on society and business.