Sharing learnings and ideas as I navigate the investing universe. Tweets not financial advice.
Joined April 2022
- Tweets274
- Following1.8K
- Followers125
- Likes453
TokenTurtle 🐂🀄️ retweeted
🛢️ The situation of AI frontier labs reminds me of the oil industry in the early 1970s. I believe we’re about to go through a similar upheaval of the entire value chain.
1️⃣ As a reminder, the starting point was that the global oil industry was dominated by the so-called Seven Sisters, Western behemoths that were fully vertically integrated, from extraction to refinement to retail distribution.
But then oil-producing countries, including Venezuela, Saudi Arabia, Algeria, and Libya, for a mix of financial, political, and geopolitical reasons, increasingly took control of and nationalized their oil reserves and production, thus breaking the value chain in two.
2️⃣ Today, there is no such thing as vertical integration per se in the AI industry, obviously. However, you could say that circular funding between Nvidia, hyperscalers, and frontier labs is de facto vertical integration. Capital circulates in a closed loop, and this makes the industry vulnerable to a disruptive event, as happened for oil in the 1970s.
3️⃣ Another thing that echoes the 1970s: we’re building too much, too fast, assuming demand will always go up and to the right and will be indefinitely solvable. In the 1970s, the industry had invested in ever larger oil infrastructure and built the likes of Batillus-class supertankers.
Today, we assume that the growth of AI consumption will continue, and we’re building ever-larger data centers. The risk is that those, just like the Batillus-class supertankers back in the day, become stranded assets due to demand destruction (whether we consume less AI or the entire industry moves toward running cheap local, open-weight models).
Here the risk lies with the hyperscalers, who carry the CapEx ($400B+/year) while frontier labs only rent compute. So the stranded-asset risk is borne by the cloud providers, but the circular funding makes the whole value chain vulnerable.
4️⃣ One thing seems increasingly certain: compute prices will keep rising because of shortages of power, memory, and copper. Robert Friedland, on the Odd Lots podcast with @TheStalwart and @tracyalloway, recently said that we need to mine as much copper over the next 18 years as we mined over the past 10,000 years, without even accounting for the electrification of the economy.
This is not sustainable, and rising input prices will inevitably be reflected in the price of cloud-run tokens. This is the equivalent of the 1973 OPEC moment for AI.
5️⃣ This, obviously, is a mortal danger for frontier labs, even more so if Anthropic and OpenAI want to join SpaceX as public companies. To avert being knocked down by price hikes and the resulting demand destruction, they must achieve asset-light downstream positioning before that price hike hits.
In other words, they must make the same pivot that many Western oil companies did in the 1970s: Halliburton and Schlumberger by specializing in oil-field services, and Shell and Chevron by specializing in refinement and downstream marketing and distribution.
6️⃣ Post-disintegration, we’ll have three categories within the new AI value chain: Aramcos (hyperscalers owning compute), Schlumbergers (service and tooling platforms, such as Hugging Face, now acquired by Nvidia), and Shells (frontier labs running model-to-customer).
7️⃣ What about the infrastructure, post-compute shock?
Training and inference require different physical infrastructure. Training clusters, which are gigantic and resource-hungry facilities, tolerate remote locations for the sake of accessing cheap power. As for inference, however, it must sit near users for latency; fortunately, they’re more like traditional data centers and can effectively be located in dense urban areas.
In terms of an analogy, trained model weights are the tanker cargo: compact and transportable. Inference facilities are the refineries: close to the customer, processing raw inputs into differentiated products in real time.
8️⃣ Why does the distinction between AI factories and inference clusters matter? Because of the political backlash.
US data center opposition has blocked or delayed $130B+ in projects in a single quarter. Training mega-clusters has become increasingly difficult to build onshore: the ability to build giant data centers near American communities will not be coming back anytime soon, given the political climate. Only inference facilities can readily be located on US soil because they blend into existing urban infrastructure.
9️⃣ So where will we be building AI factories then? One scenario is that US allies overseas, such as Gulf countries, the UK, and Mexico, host the training infrastructure. What will effectively be sovereign data centers, comparable to the oil reserves of the 1970s, will absorb the training layer: sovereign capital, cheap energy, and less permitting friction. Still, subsea cable chokepoints, analogous to Hormuz and the Red Sea, will mirror the oil vulnerability in a digital form.
🔟 Finally, if compute transactions settle in USD stablecoins backed by Treasuries, global AI demand creates a permanent bid for US government debt. Petrodollars become computedollars, and we have a new recycling loop designed to prop up the US Treasury market and keep funding the rising US national debt.
And since the value chain will have been broken in two, with the end of circular financing, we will need a commodity market for trading compute, with the emergence of a new generation of Trafiguras and Glencores, which I think will eventually become the most profitable sector across the AI industry.
Cc @mariekeflament @izakaminska @kofinas
TokenTurtle 🐂🀄️ retweeted
Late-night thoughts...
Everyone has an opinion on what to buy, yet almost nobody thinks nearly as hard about how much to own.
Some years back, I do not remember if I read this or someone told me, but it was one of those things that instantly made me pause and became ingrained in my head:
“If you want to make big money trading, you need to be a stock picker who quickly becomes a portfolio manager.”
Read it calmly and digest it, because crypto culture disproportionately worships the stock picker, yet a picker will never make as much money as the manager.
How do I understand this?
Being a picker means having the balls to enter a trade when the information is at its worst, which is also where I think most of the luck factor sits.
The portfolio manager’s job begins after that, once information starts coming in and either proving or disproving what you originally thought. At that point, the job becomes deciding whether the position deserves more size, less size, or no size at all.
Crypto mostly plays this in reverse.......People go in extremely hard at the beginning, when information sucks, because they want the home run, but then refuse to go hard once information starts proving the thesis because “it already ran.”
Which is just funny to me, and has become even funnier lately with all these “celebrity traders,” because it does not matter how many times you were right.
You get paid according to:
how right you were × how much capital you had behind it.
To me, a trader who identifies ten 100x coins but never sizes them will never outperform someone who identifies one and, when the thesis proves itself, has the guts to put 30%, fuck it, 50% of his book into it.
Finding something early is obviously valuable, but if your conviction never evolves with the information, then you are basically treating the first decision as the only decision that matters... which is stupid.
Put simply, I think the second and third decisions in trading are more important than the first.
In my head, trading is basically:
Discovery → Allocation → Harvesting
Discovery has the largest luck component. Did this opportunity ever enter your field of vision early enough?
Allocation is the ability to recognize that the opportunity is becoming less uncertain and deserves more capital.
Harvesting is the ability to translate an increasingly valuable position into realized wealth without destroying all of its remaining convexity.
Crypto spends an absurd amount of time celebrating discovery, when allocation and harvesting are where the kind of fuck-you money that kills your bloodline's work ethic is made.
Kind regards,
Your favorite Plumber.
TokenTurtle 🐂🀄️ retweeted
anything/anything
memes were solved. so i took a break. but always kept a pulse on the market. robinhood chain + wallet rollout is what first piqued my interest. saw cashcat, bought some but it wasn’t really new. we’ve played this game before. something becomes culturally relevant, someone makes a meme around it, a market forms around the attention.
then i kept looking and found something weirder: memes paired with stocks. robinhood, one of the biggest retail trading platforms in the world, had started putting stocks onchain. rwas never made sense to me, btw. i never understood why i should get excited about real estate onchain when i can just… buy real estate. but this clicked. money had already become internet-native. now real assets were becoming internet-native too. which meant they could interact with all the weird things the internet had invented around money.
and stocks are a particularly good place for this to start, because stocks are already internet culture. people have relationships with companies, tickers, founders. they argue about them, meme them, build communities around them. now the assets themselves are entering the same environment as the culture that surrounds them. i’ve been calling this culture/capital: serious assets learning to spread like memes, culture becoming a distribution layer for capital.
so i started digging into who’s building in this new game. most of what i found were launchpads. you can create memes paired with stocks. at first i thought the interesting thing was simply that you could pair the two. it’s not. the interesting thing is what becomes possible once the relationship exists. a cultural asset can now have a direct economic relationship with a productive one.
take carry, one simple consequence. a meme is basically a carryless cultural asset. you hold it and your outcome depends almost entirely on what happens to the meme itself. attach an underlying and suddenly the cultural asset can accumulate something outside itself. meme + underlying introduces carry into an otherwise carryless cultural asset. you could say this is just reflections, or cashback, or dividends, or yield farming reinvented. maybe. finance has been recycling the same primitives since bartering. the interesting part was never the primitive itself. it’s what we’re starting to attach the primitives to.
and that’s where it started getting much weirder for me. culture has always produced value. it just couldn’t keep it. we make songs, images, characters, stories, jokes and when they get valuable, something else has always had to keep the proceeds. the creator, the label, the company, the platform, the ip holder. culture spreads, institutions retain. memecoins gave culture a price, but the economic output is still mostly a closed loop of attention into price and back into attention. what happens when the cultural object itself can start keeping some of it?
memecoins already made the first jump: a viral thing on the internet could become an asset. now the asset can start owning things. what happens when a meme develops a balance sheet? when something that used to only carry meaning and attention can start owning assets and accumulating value outside itself? that’s already a strange new object. not autonomous necessarily. owning something and being able to act on your own are two different things. but something that used to only be owned is now entering the class of things that can own. things can get absurd really fast.
and this is where i think it gets much bigger than memes paired with stocks. creators have always tried to create something valuable. creating something capable of accumulating value is a different project entirely. the question stops being “how do i monetize this thing?” and becomes “what can this thing own? what can it keep? what happens to it after me?”
culture always spawns from somewhere, too. someone made the first image, wrote the song, created the character, started the joke. which unfortunately brings me back to nfts. i increasingly think they arrived before a lot of the things they were actually useful for. yes art and collectibles are real. but the internet got incredibly good at copying and distributing things and incredibly bad at remembering where they came from. when culture couldn’t keep anything, forgetting where it came from was mostly a cultural problem. when things can start owning things, “where did this come from?” becomes an economic question too. an internet full of remixable, economically capable objects is going to need a way to remember where things came from.
and then a bunch of things we’ve spent years experimenting with started looking different to me. a meta arrives, everyone loses their mind, the trade gets crowded, it dies and we move on. but the capability doesn’t die with the trade. nft mania came and went, provenance remained. defi summer came and went, programmable finance remained. the ai agent trade came and went, but software being able to hold a wallet and act economically remained interesting. memecoin mania will eventually come and go, but culture becoming instantly liquid isn’t going away. rwas might stop being a narrative one day, but the world’s assets being internet-native will remain.
the metas die. the capabilities accumulate. eventually the accumulated capabilities start to compose.
culture/capital is the version i’ve been chewing on, but the more i look at it, the more it feels like one instance of something larger: anything/anything. once all of this lives in the same environment, arbitrary things can start having economic relationships with other arbitrary things. culture can own capital. capital can distribute through culture. one cultural object can own another. the number of possible relationships explodes.
i don’t know what the final object looks like and i think pretending to know would miss the interesting part. we spent the last decade building all these weird toys separately. now they’re starting to compose.
TokenTurtle 🐂🀄️ retweeted
"but all of these are cool in the context of RWAs. crypto is just the back end, SOVs are really struggling with a lack of identity."
disagree heavily, social trading works because there is some intangible value assigned to traders that have a good track record & ppl trust their opinions, tokenization of that value is very unique to crypto
also memecoins are the primary assets w/ this social trading realm atm, & they derive all of their value from how much attention they are able to generate, they aren't RWAs at all
24/7 perps on stocks, prediction markets, and social trading are all sides of the same coin, crypto is very good at enabling speculation but also providing social proof for talented speculators
if you believe entertainment finance & the creator economy is going to 100x from here then crypto is by far the most exciting bet to make in that space & it isn't particularly close, there are very limited liquid assets that you can buy to get exposure to this atm
believe that the current moves by @RobinhoodCrypto & @X confirm this thesis even more, @RobinhoodApp sees the value of 24/7 access to *all* markets on-chain including RWAs is why they are devoting so much time to building their L2 with Arbitrum, & @X is investing heavily in the creator economy vertical with their focus on @XCreators & @XMoney + highlighting tickers in app through search and adding in subscription tiers for privileged access to creators with quality information to give out on markets
i am actually most excited for crypto than i have been in a long time because i believe this is a thesis that institutional capital can easily understand and also a thesis that aligns heavily with the younger generation
crypto as an industry will benefit from the intersection of social and finance more than anything else
TokenTurtle 🐂🀄️ retweeted
@osf_rekt is excited about 2 things:
1. $REKT buyback with @RektMkts Pro subscriptions revenues (250x$800 - fees and taxes)
2. To see what happens when they turn on monthly subscription for RektMkts and use those revenues to buy back $REKT
His ambition is to scale RektMkts to 10k users. That would create a pretty decent monthly revenue stream to buy back $REKT, not to mention the exposure it would bring to the Rekt brand as a whole.
At some point the REKT-flywheel will spin at an unstoppable pace and many faders will be left in shambles.
Get your spot in the waitinglist to enter @RektMkts
mkts.rekt.com
TokenTurtle 🐂🀄️ retweeted
What Ansem is doing with his token probably reinvents how airdrops are done forever. In hindsight it was obvious, but the idea that you have an evergreen recurring airdrop is actually the way to do it and is the highest EV way to distribute tokens.
TokenTurtle 🐂🀄️ retweeted
Zcash's new Ironwood pool is being formally verified to rule out all undetectable counterfeiting bugs, up to the underlying cryptographic assumptions.
tachyon.z.cash/blog/detectin…
TokenTurtle 🐂🀄️ retweeted
Once market realize NEAR Intents are not just better crypto routing but the transaction layer for agentic business, it will get repriced higher dramatically
Most crypto x AI takes are too small.
The real shift is from humans operating businesses through spreadsheets, dashboards, and SaaS to AI becoming the interface for computing and business.
But AI needs more than payments. It need identity, discovery, reputation, dispute resolution, escrow, insurance, forwards, and markets for work.
NEAR Intents are not just better crypto routing. It's the transaction layer for agentic business.
TokenTurtle 🐂🀄️ retweeted
The Fragile Peace
The market bought a deal. The deal doesn't exist yet.
Zero oil tankers through Hormuz since the ceasefire.
Two different deals handed to each side. Lebanon bombed while the ink was wet.
The curve is pricing resolution. It's not pricing recovery.
campbellramble.ai/p/the-frag…
![trade[xyz]: A Revolution In World Finance.](https://pbs.twimg.com/media/HFQDITkaUAAuEYi.jpg)
TokenTurtle 🐂🀄️ retweeted
Rekt Energy is now available on @Amazon US with Prime shipping.
Links below 👇
TokenTurtle 🐂🀄️ retweeted
as you probably know, i was definitely in the "4 year cycle is broken" camp. i think with BTC now at 69k, 45% off the highs (and a huge underperformance vs stocks) it's quite clear i was wrong because i think that's a big enough drop for me to classify it as a cycle.
that being said, i would much rather be a buyer than a seller here, and in the context of where i think crypto can go the upside far exceeds the downside.
i'm just simply never going to be one of those people who sht themselves at the lows, because time and time again, 99% of the money i've made in my life has come from accumulating risk in big dips, and often the best trades have been the ones where i feel like i'm the only person buying and it feels absolutely horrific.
i think it's important to recognise and understand your trading and investing style, and a large portion of CT seems to have an obsession with catching breakouts. imo, this is a relatively low probability way of making good money, and i've seen very few people do it successfully consistently. even "good traders" who make money every year...the guys that killed in the 2021 or 2024 bull runs will have made many multiples more than these guys, even if they didn't have a good 2025 or 2026.
a lot of people seem to be concerned that "crypto is dead". if you've been around long enough you will have seen this story multiple times before, and it's never the case. i think there is a great misunderstanding that "crypto MUST have a use case in order to go higher" - that's not really true...my belief is price often moves first and narratives come after to match the price. the tech always seems more revolutionary at higher prices, just like how art also looks better at higher prices.
one thing for sure, i won't be fading the 4y cycle thesis again, it's clear to me that BTC is closely held by a large group of whales and if that's the narrative/playbook they wanna keep running, then that's what they gonna do and you can't fight it. what it does mean is that we will have another great big uprising and contrary to what you may think, the seeds to your wealth are actually sewn in the bear market.
think about it this way, let's say you left in 2022 after the FTX collapse...you probably didn't return to crypto until mid 2024 to start trading again, and i bet you very few people made good money from that period until now. the people who actually made the most money, were the ones who stuck around during 2023, caught trades like $PEPE, $BONK, $MOG, $WIF, $SOL...if you were around during that period it would have been very hard for you to have not caught at least one of these...and that set you up to trade from a position of strength for the remainder of the cycle.
the problem is, when you leave and when you sell, you will certainly miss the best opportunities...you're always sitting there thinking "i wish i was early"...well being early doesn't come from fkn off and thinking it's all over...sure...there's a lot of luck involved, but you have to put yourself in a position of receiving that luck which means locking in when everyone has given up, and taking risks when others are gun shy.
the best and easiest strategy is to simply DCA and accumulate spot. too many people are obsessed at catching the bottom, and are mentally weak at being able to withstand negative unrealised pnl. however, the highest probability way to make a lot of money over a longer period of time is to accumulate lower and lower and build up a nice position, so you are ready for when thinks start moving higher again, it requires a lot of patience, and it requires a lot of conviction, because everyone around you will tell you "it's too early, lower first, i'm shorting here" etc etc...but none of these points are generally relevant to your goal.
this strategy does not work if you are taking leverage, and it does not work if you are investing money you need in the short term, because it requires you to be patient and hold for long periods of time. however, i think the vast majority of people on CT would make way more money (potentially life changing for many of you) if they employed a strategy like this, rather than chasing aggressive quick wins (ie gambling).
i am very grateful for crypto because i have made enough money to change my life substantially, but the significant majority of my gains have come from accumulating risk during 2022-23 and monetising during 2024.
this next cycle will be different for me, because as you may be aware, i don't really trade much anymore and my full focus is on $REKT...but i think the same concepts apply.
over the next few weeks and months you will see many project founders leave (already started) and communities slowly die. many coins will literally never come back and will be totally gone, and then the same project founders who left will return when the good times are back and start on their next grift.
as you should have gathered, that certainly will not be me and it certainly will not be $REKT. 2026 will mark the 5th year of our community, which really truly formed and strengthened during the 2022 bear market. it is literally our bread and butter, engrained into our personalities and that is exactly why the entire project is called Rekt!
but more importantly, i think the same concepts i mentioned about trading wholly apply to being a project founder. i believe this creates a far greater opportunity for us than if we were still in a raging bull where you have to really compete hard for attention. these are the moments when everyone gives up, and you make a name for yourself as one of the few things "still going"...and actually, if you deliver positive news and create meaningful developments, people are infinitely more impressed.
creating something successful and truly inter-cycle requires trust. it requires people to believe "this thing WILL go up again when the market resumes", and this trust only establishes itself once a coin has proven its ability to bounce back and regain momentum. part of that comes from historical patterns, for coins where there are no "founders" (like BTC obviously), but for coins with founders, people want comfort that they are not going to leave and be working as hard as they can to not just keep things going, but actually excel the position.
in then last 14 years of trading markets, these are the times i've really set myself up to make the most money, and meaningfully furthering my life goals through each cycle. that's not because i am a good trader, but it's because i know i have an edge in these environments when everyone else has given up hope and i see big opportunities, with little competition in hunting for those opportunities. it is with this mindset i intend on furthering $REKT + @rektdrinks this year and i inherently believe we have a massive opportunity to stand out as something interesting in what is a terrible market.
we made massive strides last year, we've already delivered a lot in january, and i believe we have significant and sufficient momentum to make a pretty big mark this year. if the market stays this way, i believe we have a very unique opportunity to stand out, outperform and end the year making people think "fk, this was the trade all along".
markets will be markets, but i think the worst is over...the question i ask myself is not "can we go another 30-50% lower", it's the fact that i know we can go another 200-1000% higher, it's just gonna take hard work and patience.
the way i see it, is i have made money in a very specific, high probability way for many years, and if it ain't broke don't fix it. now it's time to do it with $REKT.
TokenTurtle 🐂🀄️ retweeted
THIS IS A PUBLIC SERVICE ANNOUNCEMENT
$Umbra is still at 14 Mn Mcap
For more details read
docs.umbraprivacy.com/docs/i…