The FT has reported this in a newsletter this morning:
Big Tech has eased investor concerns about its historic spending binge on artificial intelligence, posting quarterly results that surpassed expectations and showed early signs that AI is boosting earnings. Alphabet, Meta and Microsoft were the clear winners, collectively adding more than $350bn in stock market value after reporting double-digit increases in revenue and net income.
The obvious question to ask is how tech companies can already be making significant profits from AI, which profits are significantly boosting their overall share valuation, given that this technology is at such an early stage in its development?
The answer is, in fact, relatively easy to suggest. These companies, according to their own reports, are pouring hundreds of billions of dollars into AI development at present, as they see this as the next “big thing“ in tech. When you spend that much as a corporate entity, and when the dividing line between what is capital expenditure and what is revenue expenditure is so fine, then the opportunity to represent that expenditure that might, in other periods, have been charged against profits is now to be added to capital expenditure that is recorded on the balance sheet is very tempting, not least precisely because that does boost profits whilst simultaneously assiting you to meet your claim of pouring billions into investment.
I would argue that this is the simplest explanation as to how tech companies are currently profiting from AI. Simply because of the scale of their investment programmes, overheads and other costs that might otherwise have been expensed are now being capitalised, which boosts their reported profits, and so makes their shares more attractive, and so makes it easier for them to pour more money into AI, whether that is rational to do or not.
And when I talk about overhead cost being redirected, let me be clear, this covers almost every type of cost that these companies will incur, including interest, which under current accounting rules can be treated as a capital expense if being incurred to support an investment program.
Do I really, as a consequence, entirely trust the financial reporting of these companies at a time when such massive investment programmes are going on, when the expenditure in question could just as easily be written off as a cost? My answer is that I don't, and I would suggest others should be wary. These companies are currently driving the rise in share valuation in the USA, but there may be too many accounting tricks in play for my comfort. All those tricks, I stress, might be well within the currently accepted rules of accounting, but that doesn't mean to say that I am comfortable. I have serious doubts about the capacity of those rules to deliver financial reporting that is genuinely true and fair, most especially in conditions like this, and I have the right to say so.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
When Builder.AI failed this year it had too high expenses and not enough revenue.
It had done exactly this, building unnecessary internal rolls to replace 3rd party ones just so instead of a licence cost it instead had a (higher) capex because it was ‘building IP’.
Although it is true that AI can help reduce wage costs by eliminating staff it does appear that Henry Ford’s wise words have been forgotten:
“I want my factory worker to buy my cars. If they make enough money, they’ll buy my own product”
The people evaluating stock market values do not appear to be considering macroeconomics when making their estimates of future profits, which means that stocks are consistently and significantly over-valued and that a crash is inevitable.
You may be aware of Ed Zitron already Richard, though your readers probably aren’t.
I suggest this *long* read on the AI bubble: https://www.wheresyoured.at/the-haters-gui
TLDR: only one company is making genuine, actual, verifiable profits from AI, Nvidia.
That’s because Nvidia sells the chips in the AI data centres Meta, OpenAI etc are building.
Nobody else is making a profit. Some are making revenue, but it’s not clear how they are accounting for the revenue. It seems annualised revenue is the key, so basically an estimate based on their best month the past 12 months.
Also their revenue is peanuts compared to their costs (which some cases appear heavily discounted e.g. OpenAi).
All of which doesn’t begin to count the power and water consumed by AI, or it’s well documented frequency to hallucinate.
This is what I also suspect, and that the accounting is disgusing this.
Some day that capital investment might be subject to massive impairment costs when it becomes clear it cannot generate cash.
Here is a somewhat shorter Cory Doctorow piece covering a lot of the same ground:
https://pluralistic.net/2025/06/30/accounting-gaffs/
I smell another dot-com boom and bust in the offing.
Are you thinking about that well known form of accounting that enabled Enron to stay afloat so long? ‘Mark to market’ – so called ‘fair value’? As I understand it, the practice enabled firms to bring forward future profits on their books (‘Enron: The Smartest Guys in the Room’).
I think that there were some efforts to tighten up / reform this form of accounting in 2002 and 2008, but for all we know, it has been pushed back from 2009 I think to help when markets are in downturns – enabling firms /banks to once again overstate profits, claim big bonuses and defer losses.
Of course the other issue is the ratings agencies? What did they learn from the 2008 crash? Forgive my cynicism.
Your cynicism is justified.
Right now, mark-to-market will be a self justifying exercise and it will remain that way until it isn’t, probably sometime soon.
Sadly, Richard, whilst I haven’t analysed the corporate financials, I’m pretty certain you are spot on.
It comes back to an issue you have touched on previously. How much can we trust corporate financial audits? I’m not necessarily saying the auditors are ‘wrong’, however, the way tax laws, etc., have been changed to enable tax evasion, defining what qualifies as capital investment, etc., is from the neoliberal playbook and is fully exploited.
With the background above, how can Rachel (Boot on the neck) Reeves sensibly propose that more people should invest in the stock market whilst ignoring even the advice of so-called grandees that her further deregulation, etc., will be problematic? You couldn’t make this stuff up…except it is all part of the orchestrated, dark-money funded neoliberal overarching campaign.
When I was in partnership in the UK we had a change in accounting standards requiring us to treat cash in the bank as a liability on the grounds that a retiring partner had the right to be paid his/her capital share.
The whole issue of accounting standards and reality is, to say the least, murky. However, I would understand if you did not want to pursue the shortcomings in accounting standards.
I have no doubt that accounting nonsenses are being perpetrated in AI.
There is a video coming on this – probably next week. The issue is too big to ignore. We are getting crap data from accountants.
BBC Radio Scotland News, in its inimitable, deeply ill-informed tradition today, has just had discussion about a play that is due to open on the Edinburgh International Festival, on the rise and fall of RBS. An Arts ‘journalist’ was provided to give the background context, and game up with this gem: RBS “caused a global crash”. What does that mean? Does anyone know? Certainly not the newsreader in Pacific Quay, or presumably anyone in the knowledge deprived news production team; because no clarification was sought. Nor correction was provided. After all, it makes a bigger story this way. Lehman? CDS swaps, and mortgage defaults? The disarray of Central Banks? The banking regulatory failure across the world? Nope, this is all down to RBS; all done in Edinburgh, ‘just like that’. Who knew Scotland financed the global economy? Of course the hopelessly inadequate BBC newsreader did not attempt a correction; presumably because she knows nothing about the subject matter; but no doubt will be interviewing Rachel Reeves on economics and banking as if she knew what she was talking about, if the opportunity arose.
When in God’s name, are we going to be rescued from the dead hand of BBC Scotland’s cultivated ignorance; but hey, I confess they are very, very good at couthy parochialism.
Eejits, the lot of them, I tell you. Simply that: eejits.
Richard, it is slightly unsettling to discover the local vernacular fits you like a an old and comfortable pair of bird-watching shoes
🙂
How did you know what I might be wearing right now?
This ‘magic’ accounting is going to infect the next tier too. Boards will rush to replace expensive personnel with AI systems, not recognising that in the process they are firing the expertise that makes the company tick.
I’ve seen this happen many times before, when CEOs boost profits by cutting back on R&D, for example, in preparation for an IPO, trade sale, or whatever. It ALWAYS ends in tears! What makes a company viable is much more fragile than the bosses appreciate.
Correct.
What is called organisation history is lost – and with it the capacity to avoid mistakes that such history – always invested in people and not systems – brings.