← sources

Five hyperscalers spent $230.4bn in one quarter, up 118% on the year while revenue grew 12% to 30% during that same period.

source: reddit (reddit:ValueInvesting)published 2026-11-13 01:05ingested 2026-11-14 02:42

retained text

Five hyperscalers spent $230.4bn in one quarter, up 118% on the year. During the same period revenue from these companies grew 12% to 30%. Is this Math adding up is a question investors don’t seem to want to ask themselves, however the day will come that they are forced too. As investors, we must be careful to avoid dancing right up until the music stops, and be able to walk away before the crowd. This takes discipline. 

The sell-off has undoubtedly hit certain suppliers hard eg Nvidia down 25% in a month, CoreWeave 39%. However many companies with huge drawdowns such as Arm Holdings, Marvell etc are still massively overvalued, even after tanking hard. One big factor that understates the extent of this is the way GAAP accounting measures SBC. By instead using a more accurate Economic SBC approach, you calculate the cost for these companies to actually buy back the shares they have given away, in order to cancel out the dilution sbc causes. As many of these stocks have risen significantly over the last 3 years, the real cost of this SBC is much higher than reported for many (not all) tech companies. This drastically reduces real owner earnings, which makes current market PE ratio’s seem much more generous than they really are. They can avoid this cost by not repurchasing the shares, but this then dilutes shareholders, reducing earnings per share which is what matters at the end of the day, more so than overall earnings from an investors standpoint. This is one of the few tricks many companies in the industry use to make earnings seem better than they really are. Wrongfully extending the life of GPU’s on paper, to then spread out depreciation costs over a longer period than the GPU’s are actually useful for is another, and I’ll stop there. 

Nvidia's 17 Nov print will show whether the market is pricing financing or demand. Current demand is undoubtable as Customers seem willing to take on more and more debt, but demand fuelled by debt instead of real earnings, only works till credit tightens enough that it becomes unsustainable. This wouldn’t usually be a big issue for now, but with the largely uncertain business environment, growing credit stress, weakening consumer, higher inflation almost inevitable, and supply constraints in resources such as power, oil, electrical components etc, financing becoming unsustainable is almost inevitable. As we all know Neo Clouds are extremely vulnerable to this, something I believe has not been properly priced in even with the drawdown considering the likelihood of bankruptcy for many is high. If these companies are forced to slow down due to external factors out of their control, how do they keep paying these rapidly accelerating commitments/borrowing costs? 

I’ll repeat, this isn’t as catastrophic from a solvency standpoint for the hyperscalers, but everything below including oracle, it does. From a hyperscaler standpoint some may argue that earnings to sustain all this is evident from cloud growth, but we must remind ourselves that almost all of that growth is from customers that again, fund this through debt and dilution while operating huge operating losses that will pass 100 billion annually, each. Can that really continue? 

Do people agree with me on this or no? If people were to say that they have genuinely looked at the maths of this over a couple of years, and can say it makes sense, (financial commitments in comparison to future revenues/profits) I would be very surprised but perhaps I’m missing something. 

document fetched 2026-11-14 00:48 · canonical https://reddit.com/r/ValueInvesting/comments/1xuuags/five_hyperscalers_spent_2304bn_in_one_quarter_up