Great article as always Peter. SBC is an extremely important topic that is often overlooked by investors - especially in high-growing tech stocks.
On another note, I’m curious to hear your view on the current “circular” financing happening in AI. There seems to be various opinions out there on the potential implications this has.
I have been thinking about writing on the topic, but my considerations so far have been that whatever I write need to stand the test of time - in other words avoid looking like an idiot two years from now.
But for now and in this comment I can commit to the following view. The circular vendor financing is never a good sign. It complicates the network structure in the industry and makes it more vulnerable to shocks. Japan's industry is a good example where it turbocharged it in the beginning but in the end it was a liability.
That Nvidia is using its balance sheet to create some guarantees to datacenter operators like CoreWeave up to a certain threshold and then let the operator debt finance the remaining part to build the datacenter which in turn creates demand for Nvidia chips is not a bad idea - I can see why they are doing it. But it is not a risk free exercise for Nvidia.
More generally on Nvidia I would say their results are just noise. Their demand comes from hyperscalers that create this demand for GPUs based on forecasts by end users consumption of AI models. If the ultimate end user is not finding high ROI cases in AI, the end demand and subsequent demand for GPU will do down from current levels. So I would spend much more time on figuring out what decisions CTOs are making for 2026 than what Nvidia is selling and forecasting right now.
Thanks for the thoughtful response Peter. I always appreciate how you take the “test of time approach”, so I understand why you haven’t written about this yet.
I agree with your point on NVIDIA and that we need to focus on the ROI and end-user demand in order to accurately access the potential and valuation levels of NVIDIA. On the other hand, I’m also interested in the consequences and outlooks for companies like CoreWeave, Oklo etc, who have seen increasing interest from investors on what seems to be sky high expectations for future revenue. And also that these stocks are being featured in various passive ETFs.
Great article as always Peter. SBC is an extremely important topic that is often overlooked by investors - especially in high-growing tech stocks.
On another note, I’m curious to hear your view on the current “circular” financing happening in AI. There seems to be various opinions out there on the potential implications this has.
Thanks Jonas.
I have been thinking about writing on the topic, but my considerations so far have been that whatever I write need to stand the test of time - in other words avoid looking like an idiot two years from now.
But for now and in this comment I can commit to the following view. The circular vendor financing is never a good sign. It complicates the network structure in the industry and makes it more vulnerable to shocks. Japan's industry is a good example where it turbocharged it in the beginning but in the end it was a liability.
That Nvidia is using its balance sheet to create some guarantees to datacenter operators like CoreWeave up to a certain threshold and then let the operator debt finance the remaining part to build the datacenter which in turn creates demand for Nvidia chips is not a bad idea - I can see why they are doing it. But it is not a risk free exercise for Nvidia.
More generally on Nvidia I would say their results are just noise. Their demand comes from hyperscalers that create this demand for GPUs based on forecasts by end users consumption of AI models. If the ultimate end user is not finding high ROI cases in AI, the end demand and subsequent demand for GPU will do down from current levels. So I would spend much more time on figuring out what decisions CTOs are making for 2026 than what Nvidia is selling and forecasting right now.
Thanks for the thoughtful response Peter. I always appreciate how you take the “test of time approach”, so I understand why you haven’t written about this yet.
I agree with your point on NVIDIA and that we need to focus on the ROI and end-user demand in order to accurately access the potential and valuation levels of NVIDIA. On the other hand, I’m also interested in the consequences and outlooks for companies like CoreWeave, Oklo etc, who have seen increasing interest from investors on what seems to be sky high expectations for future revenue. And also that these stocks are being featured in various passive ETFs.
I think the "test of time" is a useful leash on your ideas.
Nvidia/OpenAI/SoftBank is an interesting risk cluster given the apparent threat from Google. The turns in the AI narrative are spectacular :-)