Fernando Vidal retuiteado
Joined @michaelsantoli to discuss market outlook and how the war has transformed the oil market.
m.youtube.com/watch?v=Beb5Bc…
English
Fernando Vidal
925 posts
Fernando Vidal
@fernavid
Cofounder and Chief Data Scientist at @3F_Research. Mainly trying to post cool charts. PM $FCTE $RAA https://t.co/H7PivYC5Tx
San Francisco, CA Se unió Aralık 2012
733 Siguiendo4.9K Seguidores
Fernando Vidal retuiteado
Talking oil and the need for an imminent off-ramp w/ @michaelsantoli this afternoon on @CNBCOvertime
Thank you for the conversation Mike!
cnbc.com/video/2026/03/…
English
anthropic.APIStatusError: {'type': 'error', 'error': {'details': None, 'type': 'overloaded_error', 'message': 'Overloaded'}
$25 per million tokens and the API is overloaded from too much demand...
English
Fernando Vidal retuiteado
Went to the NYSE this afternoon to talk markets w/ @ScottWapnerCNBC @CNBCClosingBell
Oil still driving everything...next week is crucial.
cnbc.com/video/2026/03/…
English
@stevehou @MetacriticCap Yeah agree these lab valuations take as a given that AI revenues are going to explode much higher. It’s a fun exercise to try to see roughly what is priced into the valuation in terms of growth, and picture what the world looks like with that amount of AI revenue in 5-10 years.
English
I don’t think we are saying different things. To expect nearly 50% annualized sales growth for 6 years straight, which is a statistically extraordinary feat leaving little room for bad execution, should you be happy with “normal equity-like returns”? Couldn’t you easily get those returns without being exposed to such courageous assumptions?
English
What multiples on a “utility company” though? ConEd ($ED) is trading at historically elevated 20x PE with the boost from AI energy demand. OpenAI is trading at 40x sales as of its last funding round. Can someone explain to me the logic of paying 40x sales for an utility company?
Chief Nerd@TheChiefNerd
🚨 SAM ALTMAN: “We see a future where intelligence is a utility, like electricity or water, and people buy it from us on a meter.”
English
Why would the stock go up 10x though? Assuming it’s valued appropriately today, normal equity-like (rf + erp ~= 10%) returns would get you reasonable multiple compression over time as long as revenue ramps in line with expectations wouldn’t it? The stock would go up more or less than that 10% each year only based on how far away the true revenue growth curve is from this base case path of course. Magical Claude illustration of the required revenue growth path attached for critique.
English
@MetacriticCap That’s only true if the stock price doesn’t go up. If the stock price also goes up 10x you’ll still get 40x sales. If stock doubles in the next few years you are still paying 20x sales when AI is supposed to reached ASI in that time period…
English
Fernando Vidal retuiteado
GPU availability continues to plummet.
New multi-year lows in availability for: B200, H100, and A100s.
Good luck getting a B200 (availability <5%).
Oil getting all the headlines (rightly), but, for markets, the dominant secular story is surging compute demand.
English
@emollick Seedance 2 being frontier on videogen (at least publicly released frontier) and being closed weights is a data point in that direction.
English
I think it is entirely possible that there will be no new frontier open weights models at some point in the near future. Counting on the Chinese AI labs to keep making their models free forever doesn’t make sense as model costs rise & the value of having a frontier model goes up
English
This is more likely a sign of capacity limitations under massive growth than a failure in engineering. They always seem to have issues when the (human) workday is kicking off on the east coast.
Gergely Orosz@GergelyOrosz
On one end, the Anthropic team is a massive user of AI to write code (80%+ of all code deployed is written by Claude Code). They ship amazingly fast. On the other hand, seeing these beyond terrible reliability numbers suggests there might be a downside to all this speed:
English
After that quick small bump higher immediately after earnings, we are at almost no change after-hours on $NVDA, the quietest AH reaction we've seen over the last 3 years.
English
@NickTimiraos @WarrenPies The effect is also visible in sectors that did not overhire in 2021. In our report we measured residuals against trendlines that excluded 2020/21 and found negative residuals even in AI exposed sectors outside tech.
English
@WarrenPies Hard for me to sort out 2021-22 overhiring from AI. 2023 seems like it could be former, 2025, latter.
English
Fernando Vidal retuiteado
GPU availability continues to plunge.
B200 at new lows...A100, H100 approaching multi-year lows.
Demand ramping…B200 rarely available in our data
English
Fernando Vidal retuiteado
Impressive! Basically everything in your post is wrong!
The few H100s that are being sold for $6k on eBay are “parts-only” meaning they don’t function, they are only for repairing other H100s.
In reality, used H100s are still selling for $15-20k+. Not bad considering the original MSRP was $25-40k depending on the exact variant!
What’s more, NVDA still makes H100s and sells them for nearly the same price it did at launch.
Lastly, I’ve included a screenshot of the H100 rental price index showing that H100 rental prices are actually growing again. We don’t have to use your flimsy math to determine their value, we can just check the market!
English
Fernando Vidal retuiteado
@LepoulpePoulpo @zephyr_z9 You can't know exactly, but for example, Google alone was processing over 1 quadrillion tokens per month just 4 months ago.
English
@zephyr_z9 What do you use to know global token consumptions ?
English
OpenRouter is responsible for a very small amount of total token traffic (1%-2% globally)
Shanghai Macro Strategist@ShanghaiMacro
I am not an AI expert. But seriously, what’s going on? Top Chinese AI models now have a combined 47% market share. Isn’t this the biggest nightmare for the American AI firms that have spent 10X more than their Chinese counterparts?
English
Fernando Vidal retuiteado
Caliban - our new AI research assistant - is now available as an add-on for existing 3Fourteen clients.
In the demo below, Gary shows how you can easily create, save and update your own custom set of charts and indicators.
3fourteenresearch.com/caliban
English
Fernando Vidal retuiteado
Our AI research assistant - Caliban - is getting better.
In the demo below, @fernavid shows how easy it is to test strategies you encounter on social media.
We will begin offering Caliban to our clients by the end of this month. For info, sign up below
3fourteenresearch.com/caliban
English
Those indices are very opaque and prone to weird spikes and then subsequent revisions. Not saying they aren’t valuable, but they measure a very noisy and differentiated set of neocloud GPU providers. Look at my recent post asking about weird spikes for example. I look at the series now, and the spikes are gone and seem to always be rewritten to show lower price levels. This data I posted is actual pricing for a single provider, AWS. Building a GPU rental price index is an ambitious but tricky task, and I am not totally sold on the SD indices telling the whole story based on what I’ve seen. Need to have a bunch of signals to rely on here.
English
@fernavid How do you source your data? Seems very different from eg the Silicon Data
English
Spot GPU prices at AWS rising across all GPU vintages for the last few months, all while the "AI-bubble" hand wringing reached a fevered pitch. Remember, some bubblers say rental prices should be in permanent decline from oversupply and obsolescence.
English
GPUs last a long time (5+ years). AI is getting more efficient over time which means older GPUs stay relevant and valuable longer. The adoption/monetization of demand is the major real AI question from a macro “is the market overvalued” question, not technicalities about GPU lifespans.
English
@fernavid Any thoughts on the depreciation lifespan debate?
English