How AI will change the internet itself

How AI will change the internet itself

I know everyone, at the moment, is focused on the Middle East, oil prices and inflation, and some are also considering the second and third-order effects on, for example, plastic and food supply chains. But while investors are distracted by geopolitical tensions, another, potentially even sharper, transition in artificial intelligence (AI) is occurring that began at the end of last year and is accelerating.

For a while, Large Language Models (LLMs) like Gemini, ChatGPT, Grok and their ilk captured our imagination. AI felt like a sophisticated parlour trick – you ask a question and receive an answer, a more efficient web search, an online adviser or even a chat companion. And even now, the majority of those who use AI are still employing it this way. But that’s all about to change.

Agentic AI and the end of the chatbot

AI’s Chapter #1 was defined by the chatbot – a tool that could help you plan a dinner or even a holiday, but couldn’t actually cook it or book it. You punch in an instruction – a ‘prompt’ – and the LLM provides a shopping list and a menu. But you still had to provide the ‘agency’ to execute the tasks. You still had to jump in the car, do the shopping, compare prices, carry the bags to the car, drive home, follow the recipe and cook the meal.

Late last year, Agentic AI emerged, representing the moment software gained the ability to execute.

Agentic AI promises to give you a ‘worker’ that does almost all of the above, order online after comparing prices to find the cheapest, while simultaneously arranging for someone to take your dog for a walk, pay your bills, do your banking, and drop off and pick up your dry cleaning.

In November 2025, the release of OpenClaw (formerly ClawdBot) demonstrated models that no longer forgot what happened three steps ago, which had previously made complex, multi-stage tasks impossible to execute. OpenClaw introduced the ‘looping model’, which stored context in external files while running continuous AI calls.

The advance allowed AI agents to run overnight, vibecode, manage emails, or even execute advanced research experiments without human supervision.

This shift is profound. AI is no longer a tool you merely talk to. AI is fast becoming a worker you delegate to. The outcome will be businesses that aren’t restricted to B2C (Business-to-Consumer) and B2B (Business-to-Business) but also incorporate B2A (Business-to-Agent).

As these agents proliferate, there’s a solid argument the digital landscape will undergo a structural shift. If the previous era was about human-centric traffic (clicking and scrolling), the next chapter will be about Agentic Swarms.

Those swarms will fundamentally and structurally increase internet traffic, creating a host of beneficiaries.

First to benefit are the Content Delivery Networks (CDNs) – global teams of servers that work together to provide fast, reliable, and secure delivery of internet content. Akamai (AKAM) is one of the oldest and largest CDNs, handling a large share of the world’s web traffic and often the go-to choice for large enterprises and media streaming services that need ‘never down’ stability. Amazon CloudFront is the CDN arm of Amazon Web Services (AWS). If a business is already hosting its data on Amazon’s servers, CloudFront provides the plumbing used to speed up the delivery of that data to users globally.

Another beneficiary will be the edge computing providers. Edge computing providers extend cloud capabilities to local, on-premises, or network-edge locations, processing data closer to the source to reduce latency, enhance security, and enable real-time applications such as AI, IoT, and 5G. The biggest are AWS, Microsoft Azure, Google Cloud, Dell, and HPE, who are already seeing a surge in traffic as agents begin ‘pathing’ through the internet in ways and volumes humans never did or could.

Down the track, and not too far away, some companies will stop designing web interfaces for humans and start building Application Programming Interfaces (APIs) exclusively for AI agents to utilise.

And a little further down the track, the framework will be laid for autonomous robots and drivers to act as the physical bodies for these agents.

Summarising the surge

The transition from chatbot to agent will change the nature of the traffic that providers handle. Indeed, they are already experiencing this change.

Browsing traffic will expand exponentially. Unlike humans who open tabs and linger, AI agents (like OpenClaw) will hit endpoints at the exact same millisecond, creating a surge of demand that requires the massive, distributed scale of an edge network to manage.

By way of example, Fastly (FSLY) has observed that automated traffic now accounts for roughly 37 per cent of its total network volume, with a staggering 80 per cent of that slice consisting specifically of AI crawlers and agents. Akamai’s latest data shows a 300 per cent year-over-year (YoY) surge in AI-bot activity, particularly in e-commerce, where agents are now performing ‘price-scraping’ and automated purchasing at a scale that dwarfs previous human-only traffic patterns.

Instructional videos on building powerful AI-driven web scrapers are also proliferating, ensuring the traffic is only just taking off.

Meanwhile, as more developers use agents to write and deploy code autonomously, the frequency of API calls and data transfers will shift from a steady stream to a hyper-volumetric stampede.

And elsewhere, the rise of Small Language Models (SLMs) will mean more AI processing moves off centralised servers and onto these edge points to reduce the (compute cost) and latency of every decision.

Are there any investment opportunities?

Short answer? Yes, dozens. 

One tension will be energy. As more traffic requires more infrastructure, that infrastructure will require raw materials, from copper to scandium. According to one estimate, by U.S. mining billionaire Robert Friedland to maintain U.S. economic growth of 3 per cent, more copper will need to be mined in the next decade and a half than has been mined in the last 10,000 years.

At the moment, AI-connected stocks are weathering a phase of creative destruction, repricing legacy companies and sifting out those that can provide the essential utilities for this new era and those that cannot or won’t.

Value will accrue first to the ‘picks and shovels’ (Infrastructure), then to the ‘Agentic Harnesses’ – the software or operating system like OpenClaw or Claude Code that wraps around a raw AI model (like Claude or ChatGPT) to make it functional in the real world.

As the AI brains or models – Claude, ChatGPT, Grok, etc. become commoditised, the real winners could be the companies providing those harnesses – those building the systems that keep the brains and models on track for hours or days at a time ensuring that instead of waiting for you to hit <enter>, the AI keeps cycling, observing its own work, decides the next step, and acting, while repeating the process until the job is done.

The Agentic utility layer providers (AKAM, NET, FSLY) essentially supply the plumbing that AI harnesses plug into. As more companies deploy these “harnessed” agents, the traffic on these networks shifts from human clicks to high-volume, automated “token consumption.”

Analysts are categorising the opportunities into three distinct pillars: Infrastructure, Ecosystem, and Governance.

Table 1. The agentic utility framework

What to watch?

Keep an eye on Token Usage Metrics. As Nvidia’s Jensen Huang noted, the new engineering standard involves massive token consumption, and this high burn rate is the fuel for the agentic boom. But I don’t imagine the usage rate will display anything but a north-easterly trajectory.

MORE BY RogerINVEST WITH MONTGOMERY

Roger Montgomery is the Founder and Chairman of Montgomery Investment Management. Roger has over three decades of experience in funds management and related activities, including equities analysis, equity and derivatives strategy, trading and stockbroking. Prior to establishing Montgomery, Roger held positions at Ord Minnett Jardine Fleming, BT (Australia) Limited and Merrill Lynch.

He is also author of best-selling investment guide-book for the stock market, Value.able – how to value the best stocks and buy them for less than they are worth.

Roger appears regularly on television and radio, and in the press, including ABC radio and TV, The Australian and Ausbiz. View upcoming media appearances. 

This post was contributed by a representative of Montgomery Investment Management Pty Limited (AFSL No. 354564). The principal purpose of this post is to provide factual information and not provide financial product advice. Additionally, the information provided is not intended to provide any recommendation or opinion about any financial product. Any commentary and statements of opinion however may contain general advice only that is prepared without taking into account your personal objectives, financial circumstances or needs. Because of this, before acting on any of the information provided, you should always consider its appropriateness in light of your personal objectives, financial circumstances and needs and should consider seeking independent advice from a financial advisor if necessary before making any decisions. This post specifically excludes personal advice.