Question everything

As intermediaries between investors seeking a reasonable return and companies promising to provide them, the job of a fund manager is often vexed. Ebullient market and monetary conditions dangle the lure of extraordinary returns, the investment equivalent of winning the lottery. Long decades of (usually) unrealistic expected economic value creation can be priced into companies in days or months and disappear just as quickly. Surprisingly, even as bond markets seek to restore some discipline to exploding debt levels and untrustworthy governments, appetite for lottery tickets remains voracious. While Firmus might have failed to create sufficient frenzy amongst investors to support a totally fanciful $45bn valuation for a business without any significant history and little in the way of discernible business operations or balance sheet, Anthropic is aiming for a valuation roughly the equivalent of the entire Australian equity market with a similar lack of history or proven business model. Looking at the history of the Anthropic valuation, kindly prepared by the AFR, it is unclear what has changed in the past 12 months to justify an increment of more than $US1.5trn in value. Nevertheless, as highly incentivised bankers spring into action on a global lottery ticket sale, cool heads aren’t exactly prevailing.

Long experience has taught us the importance of always trying to see the forest for the trees. We are yet to find the investment error (an area in which we have vast experience!) explained by our discount rate being 6.8% when it should have been 7%. Significant and permanent mistakes always mean we missed something important. As those who’ve had young children will know only too well, learning and avoiding these mistakes often means relentlessly asking the simplest questions. How? Why? As we grapple with the ramifications of AI, these basics often seem forgotten. A seemingly obvious example at present surrounds AI regulation.

While seeking a couple of trillion in market capitalisation for Anthropic, Dario Amodei is simultaneously calling for government regulation to ‘Pace the Frontier’. 

Inviting regulation is not something we see from companies every day of the week. Why? A likely answer seems to be the potential for massive liability should things go wrong. Internalising profits while socialising costs has been a core expertise of the tech sector for some time. It is why, as the charts below from Apollo Chief Economist Torsten Slok show, productivity gains in the economy have been almost totally extracted by the tech sector as price increases. Seemingly the best way to ensure AI companies consider risks around frontier model release is to ensure they retain full liability. We’d hazard a guess paying for losses from potential cyber-attacks and data breaches will result in a little more risk control than oversight from inept government bureaucracies. Similarly, as every manufacturing business in the world has spent decades trying to improve efficiency to combat the threat from low-cost Chinese competition, it seems a little rough to suggest we need all hands on deck to protect tech giants from the evil empire. While AI may deliver significant productivity gains, the assumption that these gains will be captured almost entirely through egregious pricing and excess returns for model providers such as Anthropic is far from the only possible future.

Australian equity market ramifications from AI remain centred on duration concerns for a few technology stocks and the path and profitability of data centre development. This month’s poster child was NEXTDC (ASX: NXT). In the spirit of seeing the forest for the trees, when it comes to data centres, we don’t get it. 

Using an economic value creation lens, we cannot answer the question of why investors should have an expectation of significant excess return from fairly commoditised industrial land and building assets, nor why government should see a need to facilitate or subsidise large scale investment in power hungry real estate with no obvious payoff in improving living standards for the broader population. 

AI may well have vast living standard ramifications in areas such as health; however, this seems totally unconnected to the argument for investing in sheds full of processing power. The coincident announcement that Snowy Hydro costs had blown out to at least $25bn offer no evidence Australia has a future as a low-cost power provider, not to mention the calibre of large-scale project management. Evidence on power and transmission prices to date suggests it is fanciful to believe data centres are shouldering the full cost of power provision merely because they pay 100% of the grid connection costs. For NEXTDC specifically, the travails of power provision to its M3 data centre project in Melbourne were the source of the market value decimation in October. As is often the case, the announcement raised a series of connected questions to which there are currently few good answers. 

Starting with a few facts, NEXTDC’s accounts show an asset base in property, plant and equipment and investment properties of some $9bn at the most recent balance date. Revenue and EBITDA guidance for FY27, by which time this asset base will have increased further is around $500m and $250m respectively. Even assuming that a relatively small proportion of the $9bn asset base is contributing to earnings, and depreciation costs are ignored, evidence of strong returns on capital is sparse. Valuation has been premised on promised excess returns and per megawatt economics which are impossible to validate. Disclosure on individual asset performance and trends in true unit economics have disappeared through time. 

We see almost no reason for REIT-like vehicles not to disclose economics at an individual asset level as most do. There is usually a reason if they don’t.

Suspension of the Jemena connection works in the face of community opposition (the photo below perhaps supporting their lack of enthusiasm) and concerns around the ability of the grid to supply required power are perhaps surmountable issues, although questions around penalties, termination rights and debt covenant ramifications are unsurprisingly being raised after the train has hit the wall. An equity valuation of $4.5bn versus a net asset value of $6bn simplistically suggest opportunity. Should everything go swimmingly and the company is able to access financing and eventually develop capacity in M2, M3, S4 and other sites, we then need to determine whether there is actually any excess return over the levels investors expect/demand such that the eventual asset base is worth more/less or the same as eventual book value. Much of the growth in NTA (our view of economic value creation) to date has been through investors tipping in new equity at a significant premium to book value. Given stabilised returns are many years away, financing risks high and rationale for excess return limited, a valuation of book value less some time value and risk of penalties suggest a discount to book value seems warranted. Any premium to book is based on rosy expectations of the distant future. As always, confusing the addition of large quantities of capital with the delivery of the excess return which sees it worth more than book value is a result of focusing a little more on the trees than the forest.

Tangentially, given the NEXTDC travails, it makes sense to consider the value equation for Goodman Group (ASX: GMG), the other major exposure to data centre rollout. To cut to the chase, despite some retracement in the share price, the same approach leaves a cavernous gap between current value and a level we’d see as providing decent return prospects. 

Goodman Group is a very well run and conservatively financed asset owner and developer, with significant data centre exposure (78% of current development work in progress). However, with net tangible assets of $11.79 per share at the most recent balance date, where NTA reflects market value of investment properties (given revaluation policies of all REITs), the $55.5bn of equity market value still represents an excess of more than $30bn over this book value. 

Given we do not disagree greatly with the market value of stabilised investment properties and partnership investments, the difference needs to be largely attributed to the value of property development operations, $7.5bn of the $33.4bn in asset value. While operating earnings of $1.8bn in the FY26 year support the conclusion Goodman Group is a highly adept property developer, $30bn above book value requires an expectation of large scale and highly lucrative development activity for many years to come. In oversimplified terms, should investors believe $1.8bn is a sustainable level of profit on $7.5bn of assets and the $30bn above book value is attributed against this earnings stream, investors are paying $37.5bn, or more than 20 times operating profit and 5 times book value. As property development activities are normally valued close to book value given the competitive, volatile and risky nature of the business, investors seem to believe data centre development will have decidedly different economics or those of Goodman Group will not resemble the broader industry. These are not views we share.

On AI impact more broadly, given our scepticism on the ability to identify and value expected beneficiaries, we continue to focus on the counterfactual; which businesses have assets and competitive advantage unlikely to be decimated, and valuations are offering significant insulation in any case. What seems unlikely from our standpoint is a future in which AI can replicate and construct tangible assets at much lower cost than current levels. 

While tech oligarchs may encourage us to imagine a future where robots (which will require manufacturing facilities and raw materials) take care of our every need, making food and housing abundant whilst we all spend our days on the sun lounge (not bothering to read books as AI has done that for us), such abundance will require vast manufacturing and production capability not yet in existence. 

As is the case with data centres currently, the intersection of these plans with the physical world creates massive bottlenecks where the capacity growth of the physical world cannot keep pace. The result; either significant inflationary pressures as unrealistic growth expectations meet these constraints, or more subdued growth as the physical world bites back. Skewing towards valuable, difficult to replace physical assets remains an appealing position. Absent AI developments we would remain comfortable with this positioning given the increasingly apparent pitfalls of developed economies which have relied far too heavily on financial engineering to paper over (and perhaps encourage) stagnating living standards as infrastructure and housing development wallowed relative to population levels and asset price gains were mistaken for economic growth.

Having been concerned on the elevated level of sustainable commodity prices implied in many resource valuations, sharp price moves in both commodity prices and the companies most exposed to them is changing this equation quickly. Lithium and spodumene exposed companies have seen some of the biggest reactions, however, aluminium, rare earths and uranium are also in this category. While acknowledging our long-run price expectations, set with reference to observed cost curves, are more rubbery than in older, deeper and more developed markets such as iron ore, the investment equations in most lithium stocks now look favourable through our lens. 2026 was an instructive year in that realised prices over the period were only slightly higher than our long run expectations (around US$1,300 and nearly A$2,000 versus realised prices a little above A$2,000 for 2026). IGO (ASX: IGO), PLS (ASX: PLS) and Liontown Resources (ASX: LTR) all have no net debt currently, albeit changing as they spend to add production. Enterprise values now sit at about $3.4bn for IGO, $7.2bn for PLS and $1.8bn for Liontown. 

For the sake of simplicity, should an investor want a return of about 10% on these investment values, a level materially higher than that offered by most ASX companies and without the aid of financial leverage, we’ll need sustainable earnings of $340m, $720m and $180m respectively. Historic production, for which all producers have already progressed plans to increase was respectively 352kt (25% share of 1.41mt Greenbushes production) for IGO, 880kt for PLS and 392kt for LTR. Assuming no further production increases, investors will require sustainable operating profit of under $1,000/t for IGO and PLS and under $500/t for LTR. Allowing for sustaining capex of $200-$300/t, both IGO and PLS were above this level in 2026, while LTR is still in the process of ramping production and trying to reduce unit costs. Long-winded way of saying we think IGO and PLS, at valuations below $10k/tonne of established production, offer solid return prospects on long-life mines even without production expansion. 

IGO in particular, given an operating cost advantage over around $200/t over PLS and a revenue advantage through higher grade product, requires extremely bearish assumptions to justify pessimism. We are more optimistic, with cost positioning of both the Greenbushes and Pilgangoora mines offering scope for considerable economic value creation through expanded production and the potential for periods of considerably higher prices. 

Investing in mining businesses requires an acceptance of commodity price volatility, however, when management teams adopt sensible financial leverage (not much) and focus diligently on cost optimisation, many generate very strong economic returns. We are more than happy to sit on the other side of investors more focused on short-term commodity price direction. Similarly, the propensity of many mining investors to attribute massive value to undeveloped deposits but then run a mile when the company wants to actually raise money to pursue the opportunity speaks volumes for real life behaviour of capital markets over the textbooks. Akin to AI investing, early-stage mining projects are often priced like lottery tickets. The same optionality in established mining operations is often valued far more pessimistically. Real life says earning strong returns on capital in mining is tough. Our job is to question the wildly optimistic commodity prices, development and operating costs cited in the presentations and to provide capital only when the likely returns justify the investment. A number seem to be in that zone now.

Though we remain in the sceptical camp when it comes to the proposition of financial intermediaries in both economic value creation and the facilitation of improved living standards on which governments should be focused, not everyone shares our scepticism. TAL Daiichi Life launched an initial $11.80 non-binding proposal for the 80% of Challenger it does not already own. At more than double the net tangible assets of around $5 a share, the bidder is either optimistic on the prospects for economic value creation or prepared to accept very moderate returns. The history of Japanese bidders in Australia suggests perhaps a little of both. Either way, the traditional dance between shareholders and Board always believing the business is worth more is underway. While revised capital standards allow the company to operate a more leveraged model, increased returns must come with increased risk, and the margin which the shareholders earn for standing between buyers of annuity products and the riskier investments which provide the return remains difficult to forecast. Akin to the crazy growth in mortgage lending which Basel capital standards drove through perverted banking incentives, there are plenty of questions which should seemingly be asked on the expected economic benefits of yet more financial intermediation. They won’t be.

Market Outlook

Interpretation of operating conditions is subjective. Recent sales calls from Woolworths, Coles, Wesfarmers and JB Hi-Fi point to tougher conditions for consumers under increasing pressure from higher interest rates. Blame for the inflation which remains markedly higher than desirable undoubtedly has some roots in war and fuel prices, some in excessive government spending and some in private sector businesses which have become accustomed to raising prices to protect profits rather than looking to business efficiencies. Despite the doom and gloom on declining property prices which covers the front pages, housing credit numbers are far from suggesting times are tough.

In trying to stand back and see the forest for the trees, our perceptions of profits and margins suggest the environment remains benign. Margins for most companies, including supermarkets, fuel refiners, airlines, most commodity producers, steel, explosives and a raft of others, remain buoyant versus history. Unfortunately, as was demonstrated in the last commodity trough, productivity gain is heightened when conditions get tough. We find little evidence to suggest conditions are tough for most listed businesses.

The prospect of interest rates reaching a peak would normally leave us optimistic on equity market trajectory driven by reducing interest costs, earnings leverage and improving economic prospects. Unfortunately, the valuation starting point and the dire fiscal position of governments accustomed to propping up economies with borrowed money at even the vaguest sign of weakness complicate this equation. 

Though wild valuation swings driven by ever larger quantities of short-term focused investment continue to create attractive opportunities, overall market levels suggest a high degree of caution is warranted. A vastly greater quantity of information isn’t necessarily equating to better quality information and working out which pieces of information really matter is tougher than ever. Question everything.

Managed Fund

Schroder Australian Equity Fund - PC

Australian Shares

View

Managed Fund

Schroder Equity Opportunities Fund - WC

Australian Shares

View