Alister Berkeley

Thought of the day: tech stocks that need to invest in AI now have had capex drag FcF and, in the short run, are more like industrials that, by necessity, have regular and recurring capex spend.🤔

@yupitchris Chris at times I wish she was fictional.

Sure bro… sure 😂😂😂. Also it’s spelt Labor. If you’re going to message your fictional ex-wife, get the spelling right.

@MarkDiStef Ha, thanks, Mark. For the record, she messaged me about businesses that wanted to get out of Australia. Having just seeded new businesses, I'm disgusted with the govt waste on both sides and tax policy.

Messaging my ex wife for content.

@acxjones God help me

I'm not short Xero and have no position. This is highly nuanced accounting well beyond the capabilities of standard AI LLMs. I’ve built my own proprietary Agentic Rubric at a cost of over six figures. It runs 31 specialist agents in parallel, drawing on 200+ pages of advanced accounting & finance prompt engineering, trained from 30 corporate collapses and over 1,000 reference citations spanning two decades. It also pulls live market API data for fundamental, forensic, and sentiment analysis at scale. Humbly suggest, Andrew pick another target.

Now get this kids, you might want to be sitting down or pour a drink. Software companies pay engineers in equity to build code. The equity vests. The shares are permanent. The dilution is irreversible. AI tools arrived and collapsed the cost of building the same code by 50-90%. The engineers who built the expensive version retain their equity. The shareholders who funded it bear the permanent dilution. The code amortises off the balance sheet over 3-7.5 years. The shares never amortise. They sit in the register compounding the dilution with every new grant cycle, which is now larger not smaller because the talent war for AI-era engineers is more competitive not less. The buyback is the tell. When a company announces a buyback explicitly to offset share-based compensation dilution , as Xero $XRO.AX did on the same day as results, in the same sentence, in the Investor Presentation , it is acknowledging that the equity issued to build the code costs real cash to neutralise. That cash comes from the same distributable earnings that justify the multiple. The company is spending its earnings buying back the dilution from spending its earnings building software that costs less to build every year. This is a loop. Not a virtuous one.

Think about how many companies globally carry capitalised software development costs on their balance sheets. Every #SaaS business that capitalises under IAS 38 or ASC 350. Atlassian. Wisetech. TechnologyOne. ServiceNow. Salesforce. SAP. Thousands of listed technology companies across every exchange in the world. Every one of them has a gross capitalised software balance built at historical labour rates and historical productivity assumptions, costs accumulated over years, of what AI can build in weeks or months. Every one of them has an auditor who is now, or will soon be required to assess whether significant changes in the market constitute an impairment indicator for those assets. Every one of them has a useful life assumption that was set before AI coding tools compressed development timelines by 40-75%. Almost none of them have disclosed what that productivity improvement actually means for the asset sitting on their balance sheet. Now apply the same question to every other software company on every other exchange. What is the gross capitalised software balance? What productivity improvement has the company disclosed from AI tools? What is the implied replacement cost at that productivity improvement? What does the auditor's KAM language say about whether the market change indicator was specifically assessed? Every technology company that has been capitalising development costs at historical rates while AI tools quietly made those historical rates obsolete. Most of the market has not noticed yet. $XRO.AX $WTC.AX $TNE.AX $TEAM $NOW $CRM $SAP $INTU $MSFT $ORCL $ADBE $WDAY $HUBS $ZM $SNOW $MDB $DDOG $GTLB $CFLT $BILL #CapitalisedSoftware #IAS38 #ASC350 #SoftwareImpairment #AIProductivity #UsefulLife #FairValue #ValueInUse #IAS36 #TechAccounting #ForensicAccounting #SaaSAccounting #AIDisruption #BalanceSheet #AuditRisk #KeyAuditMatter #ASU202506 #IFRS #TechInvesting #ASX #XRO #Xero #SaaS #AIAccounting #ReplacementCost #AccountingStandards #CapitalisationRate

@TwinTurboCe1ica @Will_UKing @GeoffWilsonWAM @cjoye 😂

@alisterberkeley @Will_UKing @GeoffWilsonWAM @cjoye And you thought it was just the ex-wife who was getting the 50%.

This message exchange is with my ex-wife who is a Partner at a New York law firm working in international financial markets @GeoffWilsonWAM @cjoye 🤯

Xero $XRO.AX has NZ$720.9m of software sitting on its balance sheet. Built at historical labour rates. Historical productivity. Their own annual report says AI tools let them rebuild a product that took 6 months in 10 weeks. That is a 75% productivity improvement. Apply that to the NZ$720.9m and the replacement cost of the same software today is NZ$180m. Implied writedown: NZ$541m. But here is the thing. That writedown does not happen. Not today. Not under the accounting standards. NZ IAS 36 uses the higher of replacement cost and value in use. Value in use is the discounted cash flows from 4.92 million customers paying NZ$51 a month. That DCF is approximately NZ$4.4 billion. Against a book value of NZ$720.9m. The cash flows bury the replacement cost argument completely. For the asset to actually be impaired on a value in use basis, Xero would need to lose 1.2 to 1.5 million customers. That does not happen overnight. So there is no writedown today. The accounting is technically correct. The risk is not a writedown. It is useful life compression. Xero amortises capitalised software over 3 to 7.5 years. That assumption was made when the relevant question was how long before a competitor could build something better. Digits built a competing general ledger in less time than Xero spends capitalising development costs in a single year. Pennylane built a profitable USD$115m ARR business from scratch in five years. If AI-native architecture makes existing software architecturally obsolete faster than the amortisation schedule assumes and the evidence suggests it does the useful life needs shortening. Not impairment. Just faster amortisation. Shorten the useful life to 3 years across the portfolio. The incremental annual P&L charge is NZ$103m. Against true distributable FCF of NZ$277m. That is 37% of distributable earnings consumed by accelerated amortisation. The key is whether the useful life assumptions still hold at the next audit.

When the hunter becomes the hunted - Xero (ASX: $XRO.AX) New Competitor Landscape

@PinguCapital Oh boy 🫣🤦‍♂️

@alisterberkeley Only if it’s primary source of income though, right?

Finally.

@PinguCapital professional gambling is taxable my friend

@alisterberkeley Just in time with these CGT changes.

@John_Hempton Having sat in Glencore's and BP’s offices in Singapore for months, restructuring one of their air swings on a $2bn dollar petrochemical plant. My key takeaway: No one knows what is going on.

All the people I regard as genuinely expert on oil have been (incorrectly) picking oil prices in the one-to-three month term far higher than have actually happened. I have been around markets long enough to know experts are often wrong. Not as often wrong as mug punters but still often enough wrong that you take predictions with a big grain of salt. But I have never seen them this consistently wrong. Or at least wrong so far.

$XRO.AX Why Xero’s Adj-EBITDA is sketchy🚨🔎 1/ The standard defence for adding back share-based compensation (SBC) to reach Adjusted EBITDA is always the same: “It’s non-cash.” That defence was always weak; it became indefensible for Xero. 2/ The board authorised a buyback of up to A$550 million, explicitly to offset dilution from SBC grants. The Annual Report makes this clear: the buyback is to neutralise dilution associated with upcoming FY27 SBC allocations and historical grants. Management’s own words confirm what we always suspected: SBC is a cash cost. 3/ $XRO.AX are about to spend real money, drawn from the same balance sheet that just absorbed a NZ$5.4 billion acquisition, purely to undo the dilution SBC creates. This isn’t accounting theory. This is cash leaving the company. 4/ Walk through the arithmetic: FY26 SBC expense = NZ$317 million • Buyback authorised = A$550 million (~NZ$660 million) - BC add-back in Adj-EBITDA: NZ$240m (income statement) - SBC capitalised: NZ$77m (balance sheet, amortised back later) - Total SBC cost to shareholders: NZ$317m (the real dilution figure) Xero is planning to spend roughly two full years of SBC expense to neutralise the dilution. That cash has to come from somewhere. 5/ Specifically, it comes from the NZ$554 million of free cash flow that investors have been trained to view as the “cleanest” measure of Xero’s earnings power. The practical effect? That FCF is not freely available capital. 6/ A material portion of it, potentially NZ$330 million per year if the buyback is spread over two years, is pre-committed to buying back shares. True distributable free cash flow after the SBC buyback obligation is probably closer to NZ$220–250 million. That is the number actually available for debt repayment, acquisitions, or genuine return of capital. 7/ Now layer on the R&D capitalisation: • Income statement charges NZ$267 million of amortisation on previously capitalised development costs • Xero defers NZ$430 million of new development spend onto the balance sheet Adj-EBITDA therefore gets the double benefit of: Not expensing the NZ$430m new spend, and adding back the NZ$317m SBC 8/ Result? Adj-EBITDA = NZ$757 million This is the most flattering earnings number Xero could legitimately present, and the one most prominently featured in every analyst model and market commentary. The most honest earnings number, FCF after the SBC neutralisation cost, is probably NZ$220–250 million. 9/ The gap between NZ$757m and ~NZ$225m is not fraud. Every adjustment is disclosed. The accounting is technically correct. But the gap is roughly NZ$500–535 million. That is the distance between the number the company wants you to use and the number that most accurately reflects what shareholders actually receive. 10/ This is the SBC Buyback Loop in its purest form. Companies issue shares → dilute shareholders → report “non-cash” expense → add it back to EBITDA → market capitalise the inflated number → the company then spends real cash to buy back the very shares it just issued. It’s a closed loop that transfers value from existing shareholders to employees while flattering every key valuation metric. For $XRO.AX, the loop just became explicit and undeniable. What was once a theoretical critique of Adjusted EBITDA is now written in the company’s own capital allocation policy. Investors should price it accordingly.

@Filapek 🫣

RBC Capital on Xero $XRO: "AI disruption fears overshadowed an otherwise solid result with FY26 print beating across most operating metrics as well as guide ahead of consensus. "US segment showing strong growth and the incremental step-up in investment should see the region yield results over time. "Claude-induced selloff dismissed as another market knee-jerk reaction by some, however, we understand the implications for software if aggregators live on top, and adjust our EBIT exit multiple from 15x to 12x. "FY27 guidance quantified and sees FY27/28 EBITDA +0%/+13%. Still seeing value here if Xero delivers, remain Outperform with new $130/share PT." #stockinfocus #XJO #equities #investing

@555spectrader @markbouris @cjoye @GeoffWilsonWAM And the NAIF needs to get out of its own way

@alisterberkeley @markbouris @cjoye @GeoffWilsonWAM Special Economic Zone In Darwin. Plenty of land, water and cheap energy, close to Asia. The city has plenty of spare blocks to build high rises.

Mark et al need to get across the details of the Australian tax regime before mounting arguments otherwise they risk coming across as misinformed. -> ato.gov.au/businesses-and… @markbouris @cjoye @GeoffWilsonWAM