Under-egged

📚️ PDF⏳️ 6 min 📖 6

Eggs for the win

Recently advertised. A good job with a large accounting firm. A$87,500 including superannuation (so, $78,125). 

I found this 1998 report on graduate salaries. 27 years have passed, which is roughly ‘a generation’. Graduate salaries in accounting firms at the time were A$28,000 across a decent sample size. 

If we use indexation it reveals something startling. The equivalent of $87,500 using CPI would be $42,250 in 1998 terms. That suggests that graduate salaries are now much higher than they were. Great news for young people. 

However, if I index today’s salary against the NASDAQ over that period, the equivalent salary would be $6,803, implying a near total collapse.

We can do the same using the Australian House Price index from the FRED: in housing terms you are being paid $16,252. 

There are good arguments as to why the NASDAQ and Housing might be bad measures. For example, you might consider both overvalued. Arguably though, one represents our technical progress in a generation and the other represents the cost of a thing most people have to have, somewhere to live. They are both relevant.

Still, let’s use eggs. Eggs are a better metric because it’s difficult for them to encompass technological change. A hen lays the same number of eggs as it did in 1998. Maybe there are productivity improvements around hen health and longevity, but nothing like we see in normal tech. With the help of GPT-5, then. 

So in egg terms, salaries are down about 10%, yet in CPI terms they are up 45%. I’m doing it back to front, so perhaps it’s easier to see graphically.

Your “real” salary is up 45%. It's the only thing that rose. You can buy fewer eggs, half as many houses and only 22% the amount of NASDAQ. 

Suffice to say, the CPI is one of the largest frauds ever perpetrated on the general public.

$360 billion

In a single year, 2025, the large US tech firms will spend $360 billion (A$550 billion) on capex. Close to 25% of Australia’s GDP. All AI related, the spend is either on data centres, energy for them or the GPU chips themselves. It's an enormous investment. 

For local context the M5 Tunnel, one of the larger infrastructure projects in NSW, cost A$4.3 billion. WestConnex cost A$17 billion and took 8 years.  

Things are much slower here of course. Sydney lawyers Denton’s recently laid out the pathway for anyone that dares to build a data centre in New South Wales. It looks like an absolute minefield, requiring applicants to submit a “Secretary's Environmental Assessment Requirement” (SEARS) before submitting a development application.  

“SEARS will require the preparation of a comprehensive ACHAR as a key part of the EIS”

The acronym soup unravels as follows: the ACHAR (Aboriginal Cultural Heritage Assessment Report), is part of the EIS (Environmental Impact Statement) which are all included in the SEARS. If you get all that done you can submit a planning application, which requires approval and depends on your SEARS being approved.

Suffice to say the AI race is over and Australia isn’t in it. 

All the same, the developments in the US are intensely relevant because the big tech firms would not deploy that amount of money for no reason. They know something. GPT5 already has an IQ (if that's possible) of 135. Google used to spend $500k a year on PhDs straight out of university because they could code and were clever. Now they can spin up thousands of them in minutes. 

Take this little known piece of software, once internal to Google and now public (ish).  

Nobody has heard of Jules. You only get Jules on the top Google plan, and even then you only get 300 tasks per day and 60 concurrent tasks. It takes a codebase, clones it in the cloud and then starts making changes and running tests according to user instructions. When it’s happy that it works, it sends back the final product.  

To use it is not even that exciting, you don’t see much. It goes off behind the scenes and when you come back the function that you wanted is there and working (or it isn’t), but no time is wasted. Behind the scenes you essentially get 60 programmers working on your task at once. Google uses this product internally, 50% of their code is now computer written (a number that is probably now old).

Amusingly, the Jules interface looks like something from a 1980s computer. Presumably deliberate. 

The $360 billion of course is “a bubble” and “AI is about to collapse”. Except that it probably isn’t and what they are really building is armies of knowledge workers that mean they can cut their operating expenses in half. 

Meanwhile, I’m watching the Australian Treasurer announce that Australia is “suspending changes to its building code to boost productivity”. It might work but I somehow feel like he is operating like it's 1994 and Google and their peers are thinking 2034 and beyond.

Stable Times

The rest of the world clearly sat up and took notice of the GENIUS ACT. In Japan, they seem set to shortly approve JPYC, a Japanese Stable Coin. The issuer plans to issue ¥1 Trillion over the course of three years. 

As in the US, the assets will be 100% backed by either government debt or cash in bank accounts. 

The Bank of Japan must be mightily relieved to have another outlet for its debt. They currently hold 52% of the entire outstanding amount. Japan’s total debt is ¥1.28 quadrillion (nice to have quadrillions participating in macro figures). The BOJ holds some ¥570 trillion of that so you’d have to think that getting to ¥1 Trillion in stablecoin issuance over three years lacks ambition. I suspect it will be either a lot more or close to zero and nothing in between. 

South Korea’s new President, Lee Jae-myung is very pro the sector too. Their stablecoin legislation is due to head to parliament in October. 

We probably need one more before all the dominoes fall. The world's number one economy is doing it (USA), number four economy (Japan). It’s really about China and right on cue, this appeared on Reuters.  

I would say the sources are good because it was discussed at the Chinese Cabinet meeting. If anyone can spin up a many trillion dollar asset class in a few weeks, it’s China. 

Zero Days

The most derided suggestion so far from Australia’s productivity gathering has been the suggestion of a four day week. “Slammed by Business Groups”. Personally, I consider it the most sensible suggestion and indeed the only one that understands productivity. 

A leader with real vision might say “let’s set a goal to totally eliminate work”. You know, zero days of work for anyone. It seems unlikely until you sit down and plan out how it might happen. If you do that, things immediately come to mind. Automation, machines, software, energy - how do we get our people to work on all the things we need to make zero days happen. The trains and the cars can already drive themselves. Macdonald's outlets are already totally automated in some locations (none of which are in Australia).  

Zero days = infinite productivity. I’m surprised it isn’t more discussed. Looking at the general trend from the Australian Bureau of Statistics, the number of people who work >50 hour weeks is down from 25%+ in the year 2000 to under 15% now. 

What has really happened is that people have picked up productivity and kept it for themselves. Real wages are dropping like a stone, so everyone has responded accordingly and worked less. All the while, they are using more modern tools, so produce the same output. When measured, productivity has ‘fallen’ but it hasn’t really. People just secretly work less than they did, producing the same result and keeping quiet about it. 

Good. Why don’t we just aim for zero and do the things we need to do to get there. Four days isn’t stupid; it’s not ambitious enough.

Euro-Trash

If 1 million people died in Europe I imagine we would read about it or at least hear about it. The nature of the demographic curves mean exactly that is happening. Europe loses about 1.5 million people every year now to ‘natural attrition’. Also known as a lack of babies. Overall, headcount isn’t dropping though because of the number of inbound migrants. Accordingly, there is no political crisis about population.  

Congratulations for this success are due to people like me. When I was at school we used to raise money for a charity called “Population Concern”. Speakers would attend our school with charts showing how the population explosion would eventually devour us all. It resonated too, it was all round the time of Live Aid and after the Ethiopian Famine. The marketing campaign was real, Bob Geldoff, Bono, the full works. 

I looked up Population Concern. The charity no longer exists. It merged with ‘Plan UK’ who focus on children’s rights. Arguably somewhat detached from the original mission.  

I played my part though. I knocked on my neighbours’ doors, I raised money, spoke about the bogeyman that would come and kill us. Now, 1.5 million people in Europe that might be here aren’t. 

It was my first and best lesson in the power of collective stupidity. 

Now my own children are assaulted with new nightmares of heavily marketed “things that will kill us all”. They appear through the door with a new one each day.

Further Information

Our July 2025 report to investors can be found here.