From coal dust to Compute: the new engines of growth

The world goes in cycles. In the natural world it's seasons. In the economy it's eras. The last ten years since the recovery of 2015 we have seen the build out. The numbers in Australia clearly show this cycle coming to an end.

As Matt Comyn, head of CBA, sagely said last week the next twenty years are about whether we can continue to be the lucky country, as we enter a pivotal moment when our current economic super cycle is ending. Our current prosperity started with the digital revolution of the 1990s and has been fulled by our own openness to the world and China’s rise. Our thirty year run of uninterrupted growth has come to an end.

That cycle, built on the three Cs – access to capital, consumption and connectivity – is now giving way to a more complex reality. As our GDP and likely your own lived experience shows, productivity gains are harder to come by, structural inflation is sticky and the tailwinds of the internet age have matured into headwinds. We’re navigating a transition, still shaped by technology, but now defined by scarcity, resilience and the need for a new engine of growth.

Getting more from the same input

The great hope of course is AI. Can it, will it, fix our economic stagnation? It likely will, but to whom will go the spoils is the real question. If we should have learned one thing from the internet era of the digital age that is coming to a close it is that reducing friction creates unequal outcomes. Winners take more of the pie. In the coming AI world it is easy to envision two scenarios. 

In one world, by 2028 AI has helped unlock a new wave of innovation across industries where our productivity lags as high costs of labour, low marginal productivity gains and higher costs of capital have pulled productivity gains to zero and created the risk that economic stagnation becomes our future. In the other we are all out of work and homeless so I would rather focus on the first scenario.

Think about housing. In the internet era we are leaving, it became easier to manage everything as things were digitised. Whether it was the planning approval, paying the water bill or listing it to sell, the productivity unlock was that this was faster. In the AI era, a house might be built for less through optimised material use, cost less to run due to reducing water consumption and tell you when it is the optimal time to sell. The unlock this time is that it thinks for itself. It's not about reducing friction for humans, it's about creating actions that don’t require us.

It’s early but already clear

The first rule of technology is that if the service is free then you are the product. The first principle of running a good business is to ensure that your capital expenditure doesn’t run ahead of your operating expenditure but our biggest companies are willingly investing everything they have into computing power that destroys the defensive moat they have today. So why are they doing this? Because it is an investment in tomorrow.

Whether it's Google offering AI based answers unique to your search rather than a page of web links for you to scroll through, or Suncorp using AI to automatically process claims you are owed, everyone from minnows like Telstra to giants like Amazon are trying to buy Compute like Donald Trump seeking ‘investment’ into his family businesses. Compute is simply the new commodity required to stay relevant in the AI age. Like steam power was to early railroads, Compute is the same to early AI. Unlike steam power though, which took a couple of decades to commodify, Compute is already there.

So if compute is a commodity like electricity and phone coverage this is not where the value will be found nor where will it be created. Beyond more hygiene factors like quality of output per token there is little to suggest that anyone will find a moat like Google did in search, Microsoft in software and Meta in Social. AI is an everything technology more akin to how the engine created the industrial revolution or the cell tower enabled the last re-platforming of our world where we switched from being bound to the desktop to enslaved to our mobile.

AI is an enabler. AI is the next platform.

The last platform shift unleashed everything from Uber to AirBnB, from Google Maps to Spotify. This, though, is not that. Where Uber and AirBnB worked because they took on legacy systems with friction everywhere such as finding a taxi or booking a rental property, AI will not reduce friction between those two things were using an interface, connectivity and software. It will challenge why you needed the interface and the software in the first place.

Today the web is a series of databases. In the new world, the world’s knowledge will not be found through searching it will be presented to you through prompting. Think about Google Maps but this time rather than following directions the directions will follow you. It is already here. Waiting. Predicting where you will go based on what you and others have done time and time before.

In the AI era it will be Waymo that gets rid of Uber through making the premise of needing a car in a city start to look silly. If you have been to San Fran recently you will already know that Waymo with no human drivers already accounts for more trips than Uber does. 'Look Mum, no hands!' If you are a human professional then you have to understand you are currently more akin to the Uber driver than the Waymo machine. If you are like me (be honest folks) you are used to delivering the person, brand or business to their destination through your direction. This is what is changing and changing fast. Look forward five years and, while you will still be required, you will have to think about how you will deliver the destination through a series of prompts and AI tools that self guide the person, brand or business to their destination.

Will your agent please schedule this with mine?

If AI is an everything technology then basic argument construction* indicates that it will be in everything. It will have the ability to do everything from generate a new type of OS and therefore devices. We have seen this movie before from where the horsepower of engines replaced actual horses to how the database replaced the world of paper libraries. Canon killed Kodak, Apple killed Nokia and Jonny Ive might just be about to kill Apple with his new company io. Why? Compute is not a competitive advantage. Much like an engine in a car it enables it to move faster or slower, lift more or lift less, Compute powers aspects like speed and accuracy, but it is not in and of itself anything.

The new business model is the same as the old business model. It is about finding compounding areas of return. If Compute and the ability to build your own model is simply as easy as turning the lights on, plugging in the microwave or making a cup of tea then this is a force that dramatically reshapes value. Let me use a simple analogy.

My great grandfather was a coal miner and was, by all accounts, a mountain of a man. He made more because he could dig more in a day than anyone else. Side note he also drank more by all accounts so not sure he came out ahead. He was valued by the mining companies of the day because he could dig but was also a great example for the others who dug as well. His value was his power. Today the value of most companies is in things that are more intangible than extracting bits of a coal seam. They are in a pile of intellectual property – mostly their brand, algos, their supply chain and the unique data that they hold.

So what happens to existing players when a new technology comes along like AI? It naturally drives creative destruction for those who keep up and those who don’t will disappear. AI models help companies test drugs, design materials, optimise marketing, and simulate economic policies much faster and cheaper than traditional R&D.

What it will hopefully do is unleash innovation across our economy to fix our productivity. We will enter a world that feels more akin to no-code/low-code world future where the killer skill moves again from the ability to code your idea to the ability to describe your idea. This new era could be great, or it could be shit, but its happening all around you if you care to look for it. Whether you have your eyes half closed like watching a horror movie or you are all in on the coming world of AI agents, the most important thing we have to remember is that we have agency.

What does AI look and feel like?

As the new world appears we are already seeing the brand language take shape. If the brand semiotics of the app and cloud era looked like lower case fonts, happy colours and soft forms of things, the brand language of AI is about humanising the front end of a brand new technology. Robots are only scary when you call them killer so the branding in this space has all been designed to help a technology that is alien to us feel more human.

Everything from Chat to Claude, Gemini to Co Pilot, Leonardo to Dall-e, Antrophic and Writer, are all named and branded to make us feel like we have seen them before. Language, words and symbols from our current world to brand the Large Language Models shaping our future.

There is of course another side to this – AI being used for surveillance and security where brands like Plantir, Shield AI and Proteus (Amazon) pull from darker sides of our world; and the need for protection and security is all wrapped up in the semiotics of securing our freedom and happiness.

As the old proverb goes, 'Be careful when the symbols of your past are being re-interpreted to sell you your future'.

—
Be better to each other.

*Editor's note, a rant: argument construction is the biggest failing of our current education system that focuses on encouraging consensus rather than encouraging constructive disagreement