Monthly Market Update – Aug 2026

Performance data and commentary across asset classes and strategies.

Tables are sorted by 1-month return. Benchmark rows are highlighted.

Data current as of end-Aug 2026.

What’s New

Some formatting changes to the newsletter and some data that you might find interesting.

• Mortgage ratesCurrent mortgage rates from all Australian lenders.
• Term depositsTerm deposit rates by term, amount and bank.
• Savings accountsSavings rates including bonus rates at any balance.
• Super LabPerformance, fees, asset allocation, default insurances (!) and the ability to compare any industry super fund.
• CPI contributionsMonthly CPI split into the groups that drive it. Choose a region and a horizon. Drill down to sub-groups.

01

Fixed Income

• I was trying to Hail Mary a prediction but CPI came in way hotter than I was hoping, eek!
  • Inflation is +3.45% over the year and this is as good as it gets for some time.
  • Mechanically near impossible for that number to now dip below 3.5%, 4% is the min level we hit.
• Short duration credits seemingly the only place that has beaten CPI+2% returns.

Fixed Income	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark

02

Private Credit

• Ah sooo, this sector is absolutely cooked.
• I really wish I could say more but my god are we in for some pain.
• CRE lending exposure is so incredibly challenged, why would anyone buy an apartment off-the-plan? But pre-sales needed for finance.
  • The key to watch now is apartment price levels, if these fall persistently then it jeopardises settlements.
• Rate rise is going to challenge things further.
• Not trying to be early here, but there are going to be LOTS of opportunities in market, spreads have normalised and not much extra cash floating around.
  • Challenge in how to get exposure, there are deals to be done but you don't want to be co-mingled with existing loans.
  • Be a provider of liquidity when there is a shortage.

Private Credit	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark

03

Domestic Large Cap

• Gold, healthcare and commodities. If you held these, you did well.
• Over half my list generated negative returns over 1 year despite the index being up +4.4% over the same period.

Domestic Large Cap	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark

04

Domestic Mid/Small Cap

• Lots of beta this month but little alpha.
• Regal. Phil announces retirement, so how much capital is sticky to him?
  • Trust in Phil or trust in team?
  • Anecdotally, I've known of offshore managers (n>1) where the "Phils" of those funds stuck around in a largely ceremonial fashion.
  • I'm surprised it was declared so authoritatively, why not just turn him into Howard Marks of Australia?

Domestic Mid/Small Cap	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark

05

Domestic Micro Cap

• Index just keeps ripping due to speccy gold & lithium names, no sane manager holds that much risk.
• Tax changes hit this sector the most as all gains are CGT.
• I worry about future activity if the post-tax hurdle is just so much higher.
• But while commodities/precious metals markets continue to pump, this is the juiciest sector.

Domestic Micro Cap	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark

06

International Equities

• Lots of "quality" managers getting a continued reprieve but growth came back hard as NASDAQ snapped back.
• Index has concentration risk but most funds here run as much (if not more) concentration themselves, unfortunately only one sector generating returns.
• This space is completely about sectors, themes and factors, active management be damned.
  • Rather than picking a stock, much easier to just pick the sector and theme that is running.
  • The passive-ification of investing is wholly linked to a particular factor working really well - size!
  • Something I think most have forgotten, that size itself is a factor and one that has worked well through cycles, this is not an aberration.

International Equities	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark

07

Infra + REITs

• Not a single positive number for the month but infra 1 year numbers still intact.
• A battle between the datacentre/electricity bet and where the 10YR rate is heading.

Infra + REITs	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst

08

Other

• GDX +30% for the month! Gold miners have compounded at 47%pa for 3 years and surprisingly not much exposure to them relative to something like the AI play.
• Trend (Aspect, AQR, Winton, PIMCO) keeps absolutely smashing it. Long gold, long memory, short bonds. Machines get to the conclusion faster than us mere mortals and a volatile macro backdrop always good for these.
• Top of the table for 1M dominated by funds that could hold dirt. Inflation highly correlated to commodities + trend (AQR paper is a must read).

Other	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst Benchmark (Spot Gold)

09

GICS Sectors

• Healthcare +19% for the month but still down -18% for the year.
• Materials +12% for the month and +49% for the year.
  • Imagine being an Aussie and trying to avoid owning the miners.
• Discretionary and A-REITs reflective of consumer and property getting hammered by rising rates. Two-speed economy in a table!
• Among the wildest dispersion I've seen on the ASX in a while.

GICS Sectors	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa	10 Yr pa

Best Worst Benchmark

10

Crypto

• Here we go!
• A 22% bounce off a 50% drawdown is still a 40% drawdown. Brutal.
• Long way to go to get back to ATH, opportunity remains.

Crypto	1 Mth	3 Mth	1 Yr pa	3 Yr pa	5 Yr pa

Best Worst

Thoughts of the Month

AI economics, the competition for capital, and safety

AI

Random thoughts of this month:

• My token usage is going up exponentially but my token cost is going down, net monthly cost is lower than ever.
  • Makes me wonder how persistently OpenAI/Anthropic can generate $$$ in the future when the hurdle to switch is so so low?

• Speaking of $$$ and profitability, below excerpt from Semi-Analysis piece talking about Google's accelerator chip "Ironwood".
  • "...At an interactivity of 100 tokens/s/user, Ironwood costs approximately $0.181 per million total tokens...".
  • That cost is all-in datacentre cost (chip and server capex, networking, datacentre hosting, power and cost of capital) to serve a large model.
  • For reference, Claude Fable 5.1 costs $10 per million input & $50 per million output tokens and is hosted on the same hardware (Ironwood).
    • The all-in cost to serve this model (inference) would be higher than the $0.181 quoted above, but even if it is multiples higher, it is minuscule relative to implied revenue.
  • Inference margins are over 90% for the AI companies - ie the cost of serving a model is so incredibly profitable it is mindblowing.
    • Yes, training and research is where all the costs lie but the marginal token unit economics (ie a new user being acquired) are extraordinary.
  • If you're wondering where the AI growth is coming from and what is justifying it, here it is, it's the absurd margins. Datacentres earn infra returns and if you can serve a model on top of this infra, then the margins explode. Also the reason why everyone is scrambling to secure their own power and chips.

• META going to come out with MUSE in Australia soon, autonomous end-to-end agent for personal use meaning you can tell it to go do something behind password walls and it will execute it on your behalf.
  • I'm less interested in that specifically (OpenClaw/Hermes has existed for a while) but more interested in a similar framework for enterprise, hasn't been nailed yet by anyone.
  • I have ideas for financial services, if you want to chat please reach out.

Pouring Fuel on Fire

Goldman put out a note last week called "Competition for Capital" in which their argument is hyperscalers are spending so much on AI infrastructure that it's eaten up free cash flow and leading them to tap both debt and equity markets. Governments at the same time are borrowing for defence, energy security and infrastructure. We've gone from the post-GFC excess capital to everyone coming to the dinner table at the same time, and the price of money has gone up.

Which got me thinking about a question that sounds stupid when you say it out loud.

Do higher rates cause inflation?

Textbooks obviously say no as when inflation goes up, the central bank lifts rates, money becomes more expensive, demand cools, and inflation comes down. And you keep raising rates until inflation is tempered aka the beatings continue until morale improves.

Earlier this year I wrote about personal investment hurdle rates but it's now worth covering another hurdle rate channel.

Higher interest rates means a higher weighted average cost of capital (WACC). A higher WACC means a higher hurdle, and a higher hurdle means it's harder to justify capital investment meaning fewer projects get built. Fewer projects means less supply, and if demand hasn't gone anywhere, less supply means higher prices. The circle of life for capital markets!

We see this in mining all the time. The <insert commodity> price needed to justify a greenfield project at a 12% discount rate is a very different a price needed at an 8% discount rate. Even with the same orebody, same capex, same strip ratio, if you change the cost of money the incentive price moves too.

So rates up, inflation up?

I haven't answered the question yet, I know

Capex boom pushed up rates, rates kill supply, lack of supply pushes up prices, prices keep rates high, repeat until morale improves. Wrong.

Rates are high because of the boom. If enough projects get cancelled, demand for capital falls and so does the hurdle rate. A rising cost of capital is merely a rationing mechanism. A mechanical method of working out what project gets the money when there isn't enough to go around. Darwinism.

A cancelled project means less demand for inputs. And any supply that doesn't get built only shows up as a problem three to five years out, ie supply is constrained and prices respond. Disinflationary now, inflationary later.

So who gets the short end of the stick?

Definitely not the Hyperscalers!

Today, a hyperscaler could not care less whether its hurdle is 8% or 12%. This is the race for the most valuable technology since electricity and the margins justify the capex spend. Governments are even worse, nobody in Canberra or DC cancels a submarine because the 10YR moved +100bps (or +200bps). Nor should they, investment is required.

The fundamental problem is the two biggest bidders for capital are the least sensitive to the cost of capital. There is no transmission mechanism here at all.

That leaves everyone else to cop it! The apartment developer whose feaso stopped stacking up at 10% borrowing rates and can't access bank capital. The transmission line earning a regulated return. The small industrial business that wanted to borrow for a second site. Any kind of low return, long-duration stuff that makes up a chunk of the actual economy.

How about the family with kids and a mega mortgage in Sydney?

When the RBA or the Fed tightens, they're trying to cool inflation and demand. But if you're battling a dual issue of supply-side energy costs rising (no view on whether the war persists) and demand-side rate insensitivity, then the entire burden to control lands on sectors that were never the problem in the first place!

We are going to end up fighting rising energy prices and an AI & defence boom by shutting down housing construction and consumer expenditure.

So What

Higher rates don't cause inflation but higher rates are going to have no effect if any on inflation today. Sure the Iran War dies down and energy prices normalise, but the structural capex boom and higher rates leads to a higher cost of capital and leads to the wrong people suffering.

Higher rates aren't going to solve supply side issues, probably make it worse. So maybe higher rates then do cause persistent inflation? Dumb question with a dumb answer perhaps.

There is a happy ending here though! Railways, electrification and fibre were all inflationary while being built and deflationary once up and running. If the AI buildout is more or less the same, it represents capital investment for future productivity. Expensive bit before the cheap bit.

AI Safety

• There has been a lot of noise (specifically by Anthropic) about AI safety and rogue models.
• It's even made it into the headlines in the AFR, that AI will kill us as researchers have given arbitrary numbers (10%?) on the end of civilisation.
• I've always thought it is hard to have a sane conversation about this.
  • If you don't agree are you being lazy?
  • If you agree with the end of times, then are you being crazy?
  • Either way, you either sound dismissive or like a nutjob.
• I take the view a more sinister threat is in humanity slowly slipping into a doom loop. A Brave New World perhaps.
• I couldn't care less about violent catastrophes, people care too much to stop that from happening.
  • My biggest worry about AI safety and rogue models is a Huxley-like alignment failure which ends up being a slow & comfortable (and voluntary) trap for humanity.
  • We won't ever know it led us astray.
  • In a few years, we are going to be too intellectually dependent to notice to correct the error.
• Countries are constantly at war with each other, mostly non-kinetic, but everyone is acting against each other in some way or fashion.
• These LLMs have so much influence over what we do, so messing up alignment is another attack vector.
  • Let alone using models from a known aggressor.

Get the next one.

Subscribe by email ↗

---

Home · Newsletter · Essays · Lab · Super · Consulting · Contact
Work with us: Fund Inclusion · Sponsorship
© 2026 Ausyield