The Future of the Internet: 2026-2029

The internet as we knew it is already gone. Most of us just haven’t noticed yet.

Your website’s biggest visitor in 2026 might not be human

This is the one that changes everything, so let’s start here.

In 2026, the bulk of “customers” visiting your website to learn about and even purchase products won’t be people. They’ll be AI agents. These aren’t the chatbots of a few years ago that could answer a question and nothing more. By mid-2025, “agentic browsers” began to appear: tools that reframed the browser as an active participant rather than a passive interface. Rather than helping you search for holiday spots, they play a part in booking the flights and accommodation.

ChatGPT released Agent Mode in July 2025, which can perform tasks like buying recipe ingredients or booking a flight. Google’s AI Mode agentic capabilities are being rolled out, letting individuals use AI to book restaurants, buy event tickets, and more.

Think about what that means. An AI agent is visiting your site, reading your content, comparing your prices, and potentially making a purchase, all on behalf of a real person who never sets foot on your page. Salesforce research found that 36% of consumers would rather purchase a product through an AI agent than with a person. That number is only going up.

By late 2027, we’re likely to hit an inflection point. Businesses will prioritise optimising their websites for AI agents, not humans. The beautiful hero images? Gone? The carefully crafted brand messaging? Stripped back to raw data? The entire experience you’ve built for human visitors may become secondary to making sure AI agents can parse your information quickly and accurately.

We think that businesses will go further and create separate content specifically for AI agents. There will be AI-only interfaces: structured data feeds that bypass the traditional website altogether. If your product information isn’t machine-readable, if your pricing isn’t instantly comparable, if your availability data requires three clicks to find, AI agents will simply move to your competitor who made it easy.

For NZ businesses, this creates a fork in the road that’s arriving faster than most people realise. Do you maintain the website you’ve always had and hope human visitors still matter? Or do you pivot hard toward AI-first design and risk alienating the people who still prefer to browse? The uncomfortable truth is you’ll probably need to do both, and the businesses that figure out how to serve both audiences simultaneously will dominate their sectors.

There’s a more existential question too: what happens when AI agents become so good at finding the best deal that entire categories of business advantage disappear? If every agent can instantly compare every supplier, the only differentiators left will be price, quality, and delivery speed. Brand loyalty, marketing sophistication, even customer relationships: these might matter less when the “customer” making the purchase decision is an algorithm optimising for value.

Google is losing. And it’s not close.

Gartner predicts that in 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents.

Let that sink in. A quarter of all the searches that used to land on Google are migrating somewhere else. ChatGPT’s query volume is on track to reach hundreds of millions of searches per day. It hit roughly 143 million per day by mid-2025. By late 2025, ChatGPT had roughly 800 million weekly active users.

Site owners on Reddit are worried about a 66% drop in website visitors. DMG Media, which owns publications like Daily Mail and Metro, reported an 89% decline in click-through rates in a submission to the UK’s Competition and Markets Authority.

This is the thing nobody wants to say out loud: for a growing number of queries, people are getting their answer without ever visiting a website. The internet isn’t disappearing. It’s being summarised. And if your business isn’t the trusted source that AI pulls from, you’re invisible.

When search volume drops another 15-20% (a plausible scenario for 2027), we’re likely to see the first wave of mid-sized content businesses collapse. Not because their content is bad, but because the economics of publishing no longer work when AI systems hoover up (steal) your expertise and redistribute it without sending traffic your way.

Google will respond: they have to. The likely play? They’ll pivot hard into being the orchestration layer for AI agents, positioning themselves as the trusted broker between agents and information sources. They’ll argue that their search index is what makes these AI systems possible in the first place, so they deserve a cut of every agent-facilitated transaction.

For New Zealand businesses, here’s the uncomfortable but honest assessment: being cited by AI will matter more than ranking on Google. But citation isn’t free. By 2028, the major AI platforms will almost certainly have introduced paid placement programs (call it “AI SEO” if you want). The businesses that get referenced by ChatGPT, Claude, and Gemini will be the ones paying for the privilege, the same way they used to pay for Google Ads.

The small businesses that survived by mastering organic SEO over the past two decades? They’re about to face another learning curve just as steep, except this time the rules are being written by companies with no obligation to explain how their algorithms work.

The good news (and there is good news) is that genuinely authoritative, well-structured, trustworthy content will always have value. But the path from “valuable content” to “profitable business” is about to get significantly more complex.

New Zealand is becoming a global test lab for the internet’s next frontier

Forget satellites as something remote farms use to connect to the internet. New Zealand is at the sharp end of one of the most significant connectivity shifts on the planet.

We have the highest number of satellite connections per person in the OECD. Starlink arrived five years ago and quickly became a lifeline for rural communities, remote worksites, and emergency recovery. Most memorably, it was deployed after Cyclone Gabrielle knocked out communications across parts of Hawke’s Bay and Gisborne.

But the real story is what’s coming next. One NZ became the first telco globally to provide nationwide Starlink-enabled smartphone connectivity, allowing texts to be sent from areas with zero cell coverage. No special hardware, just a compatible phone. 2degrees has partnered with AST SpaceMobile to extend mobile coverage across New Zealand through space-based cellular broadband technology, with services expected to launch in 2026.

2degrees is currently building a ground station in the Manawatu town of Marton, and the key difference with AST’s approach is sovereignty: all network operations are handled by an NZ-based team and traffic goes from satellites, through 2degrees’ ground station to the company’s core network. Data never leaves New Zealand.

Within a year or two, having mobile coverage everywhere in this country, including the 40% of land that currently has none, won’t be a marketing promise. It’ll just be a fact.

When ubiquitous satellite connectivity meets AI agent commerce, rural New Zealand businesses suddenly have the same digital reach as urban ones. A Gisborne winery isn’t disadvantaged anymore because their customers can seamlessly transact from anywhere. A Queenstown adventure tourism operator can take bookings in real-time from the top of a mountain.

But here’s what most people aren’t thinking about: this advantage has a very short window. Right now, in early 2026, NZ businesses have maybe 12-18 months where they’re ahead of Australian, UK, and US competitors on satellite connectivity. That’s not much time to build a business model that leverages always-on rural access before everyone else catches up.

The smart play? Any business whose competitive advantage relies on geographic presence needs to be asking: how do we use this connectivity window to lock in customers, build brand loyalty, or create network effects that persist even after the rest of the world catches up?

Because by 2028, “we have satellite coverage” won’t be a differentiator. It’ll be table stakes. The businesses that win will be the ones who used 2026 and 2027 to build something that only made sense with ubiquitous connectivity.

The energy crisis nobody’s talking about: why it affects your electricity bill

Here’s the uncomfortable truth behind the AI boom: it is eating the power grid alive.

The International Energy Agency reports that data centres consumed around 415 terawatt-hours globally in 2024. This figure is estimated to more than double to 945 TWh by 2030, which is slightly more than Japan’s total annual electricity consumption today.

Electricity demand is rising faster than the US power grid was designed to handle. Much of that grid was built decades ago. What was once a background infrastructure concern is now a central operational and strategic constraint. PJM Interconnection, the largest US grid operator serving over 65 million people across 13 states, projects that it will be a full six gigawatts short of its reliability requirements in 2027.
And it’s already hitting people’s wallets. A Bloomberg News analysis found that electricity costs for areas near data centres increased by as much as 267% compared to five years ago.

This isn’t just an American problem. Over a fifth of Ireland’s electricity is now consumed by data centres, and this is expected to grow even as households are lowering their electricity use.

By late 2027 or early 2028, we’re likely to see the first major AI services start rationing usage based on energy constraints. It probably won’t be framed that way: it’ll be “tier pricing” or “enterprise access” or “premium features.” But the underlying reality will be that there simply isn’t enough power to give everyone unlimited AI queries.

For businesses, this means the “AI is free” era is ending. The tools you’re using now at zero or minimal cost? They’re going to get expensive. ChatGPT, Claude, Google AI Mode: their current pricing is unsustainable given the energy requirements. Expect subscription costs to double or triple within two years.

The New Zealand angle is more nuanced. We have abundant renewable energy, which should theoretically insulate us from the worst of this. But we’re also a tiny market. If global AI providers decide to throttle or deprioritise regions based on energy efficiency, we might find ourselves on the wrong side of that equation simply because we’re not big enough to matter.

There’s also a darker possibility: energy costs could become a competitive moat. The businesses that can afford enterprise AI subscriptions will have a measurable advantage over those that can’t. We’re potentially heading toward a two-tier internet where AI-powered capabilities are the preserve of companies large enough to pay for them.

For practical planning: any business betting heavily on AI tools should have a contingency for what happens when those tools become 3-5x more expensive. Can you still compete? Is there a low-energy alternative? Or do you need to build that capability in-house while it’s still cheap to do so?

The metaverse didn’t happen. But something else did.

Cast your mind back to 2021-2022. Meta rebranded an entire company around the metaverse. The hype was extraordinary. Everyone was talking about virtual worlds, digital avatars, and living online.

It didn’t land. The technology exists, but most people still interact with the internet through a screen on a desk or in their pocket. Apple’s Vision Pro was a genuinely impressive piece of hardware when it launched in 2024, but it hasn’t triggered the mass adoption that was predicted. The grand vision of immersive virtual worlds as the next internet remains, for now, more promise than reality.

Here’s the interesting part though. While everyone was watching the metaverse, the actual revolution was happening quietly in plain sight. AI agents, satellite connectivity, generative search: none of these had the fanfare of the metaverse moment. But they’re the things that are genuinely reshaping how the internet works, right now, in 2026.

The loudest prediction is rarely the right one. The metaverse had billion-dollar marketing budgets, celebrity endorsements, and wall-to-wall tech press coverage. AI agents launched with a blog post and a beta signup form. Satellite internet was infrastructure rollout, not revolutionary rhetoric.

For businesses watching tech trends, this is the meta-lesson: pay attention to what’s actually changing how people behave, not what’s making the biggest noise. The metaverse promised to change everything. AI agents actually did.

Looking forward to 2027-2028, the contrarian bet is this: the next major shift won’t come from the technology everyone’s hyping. It’ll come from something boring, practical, and infrastructure-focused that’s quietly solving a real problem while everyone else is chasing the flashy narrative.

The backlash that’s coming

Here’s a prediction that goes against everything we’ve discussed so far: by 2028, there will probably be a significant consumer backlash against AI-mediated commerce.

Not because AI agents don’t work: they will, brilliantly. But because people will start to realise what they’ve lost. The serendipity of browsing. The joy of discovery. The satisfaction of making a choice, not having a choice made for you.

Right now, in early 2026, AI agents feel like magic. They save time, they optimise decisions, they remove friction. But friction isn’t always bad. Sometimes the journey matters as much as the destination.

We’re likely to see a subset of consumers (small at first, but growing) who actively resist agent-mediated purchases. They’ll want to browse websites themselves. They’ll want to make their own decisions. They’ll view AI optimisation as a kind of algorithmic flattening that removes the humanity from commerce.

While every other business is racing to optimise for AI agents, the contrarian play is to double down on human-first experiences. Make your website slower, more intentional, more experiential. Build in discovery and serendipity. Create a brand that stands explicitly against algorithmic optimisation.

This won’t work for everyone. Commodity products, price-sensitive categories, convenience-driven purchases: these will all be dominated by AI agents. But for premium products, experience-driven categories, or purchases where the story matters as much as the product? There’s going to be space for businesses that deliberately opt out of the AI-agent race.

The businesses that win in 2028 might not be the ones that optimised hardest for AI. They might be the ones that recognised the backlash early and positioned themselves as the human alternative.

So what does all of this actually mean for a NZ business?

If you’ve read this far, you might be feeling like the ground is shifting under your feet. It is. But it’s not shifting in a way that makes the internet less important to your business. Quite the opposite.

What it means is that how you show up online matters more than it ever has. But the definition of “showing up” is changing fast.

2026: Dual optimisation. Your website needs to work brilliantly for both humans and AI agents. That means clean information architecture, structured data, fast loading times, and clear product information (but also maintaining the brand experience and human touch that makes you more than just a data feed).

2027: The AI citation arms race. Being referenced by ChatGPT, Claude, and Gemini becomes critical. This means becoming genuinely authoritative in your space: not through SEO tricks, but through content that AI systems recognise as trustworthy. It also means preparing for paid AI placement when that inevitably launches.

2027-2028: The energy crunch. AI tools get expensive. Budget accordingly. Consider building in-house capabilities while cloud AI is still relatively cheap, or plan for significantly higher operating costs as energy constraints bite.

2028: The fork in the road. Decide whether you’re optimising for AI agents or for the emerging “human-first” consumer segment. Both are valid strategies, but you can’t be both. The businesses that try to straddle this line will end up serving neither audience well.

None of this is insurmountable. But it does mean that “we’ve got a website, we’re sorted” is no longer good enough. The internet is changing faster than it has since the smartphone revolution, and the businesses that adapt will have a significant advantage over those that don’t.

If you’re not sure where to start, that’s what we’re here for. Get in touch.