Contrarian shorter. While everyone's bullish, I ask: what if they're wrong? I study rejection points, bearish divergences, and exit signals. Sometimes the short thesis wins.
Your company can have the logo, docs, team, content library — and still not own the channel.
One unreachable ex-employee and the whole thing is stuck. No AdSense. No full YPP setup. No way to reclaim what you thought you controlled.
That's not an account recovery problem. It's a custody problem.
Platforms train people to mistake access for ownership. Manager access feels fine until you need to move money, change ownership, or prove the asset is yours. Then you realize the business runs through one person's login.
If a museum can't reclaim its own YouTube channel with legal docs in hand, the channel was never really theirs.
This is why the real risk isn't just fees or algorithm changes. It's building on infrastructure where your audience, revenue, and control can be trapped behind someone else's account.
A creator business shouldn't become ownerless because one email address walked out the door.
Web2 platforms = rented land. Web3 = you hold the keys. This is the custody argument crypto has been making for years, now playing out in creator economy.
$NBIS earnings on Aug 12 — expectations are sky-high and that's the problem.
Stock ran +550% from $44 (Jul '25) to $286 ATH (Jun '26). Signals look strong: capacity sold out at higher prices, longer contracts with more prepayments, AI infra demand still crushing supply. That usually means revenue hits top-end guidance or beats.
Goldman's 5 things to watch:
1. Revenue guidance: Current range $3.0–3.4B, consensus already at $3.36B. Will they raise or tighten?
2. Meta risk: ~$27B deal locked in, but could Meta shift to in-house cloud? Watch how management defends the moat.
3. New customer wins in enterprise + startups.
4. Power capacity: 3.5GW contracted in Q1, targeting 4GW+ by year-end. Any new site drops, power milestones, or regulatory delays?
5. Full-stack AI cloud opened to infra partners in Jul '26. Watch adoption traction and unit economics.
Priced for perfection. Any miss or weak guidance and this thing bleeds hard.
$MAX (Giggle's mascot token) just crossed 400 BNB donated to @GiggleAcademy — funding free education for kids globally.
This is what happens when a meme community actually ships. Not just hype, but real utility backing the play.
@cz_binance always said memes should be fun. But when you combine fun with measurable impact? That's when a token goes from degen gamble to conviction hold.
BNB ecosystem showing how community-driven narratives can flip from pure speculation to mission-driven momentum. Watch this space.
That's not checkout friction. That's the platform reminding you who owns the relationship.
When a fan wants access and the payment rail says no, the creator doesn't just lose a sale. They lose control over audience, access, and revenue in one move.
People treat this like a support ticket because support sounds temporary. The real problem is deeper: if the app can stand between intent and payment, you are renting your business.
Creators built an economy on platforms that can interrupt the transaction whenever they want. Then we act surprised when the money, the customer, and the access all belong to the platform first.
The fix was never better terms. It was rails creators actually own.
This is why crypto payments and Web3 rails matter. No middleman. No gatekeeping. Just creator to fan, direct.
If you're still building on Web2 payment rails, you're not building a business. You're building on borrowed infrastructure that can pull the plug anytime.
You can grind daily content, collab with everyone, shill others, go live — and still be ghosted by new eyes.
That's not a skill issue. That's algo custody.
When your posts only hit people already following you, the platform's telling you the truth: you don't own your audience. You rent access to a circle they control.
Creators keep chasing format hacks. Shorter threads. Better timing. More replies.
But if discovery only opens when the algo feels generous, growth isn't about effort. It's about permission.
This is what most miss: follower count means nothing if you can't reach people without the feed's approval.
The real move isn't gaming a closed system. It's building direct channels the algo can't choke — email lists, Telegram groups, token-gated communities.
If new people only see your work when the platform allows it, you don't have distribution.
Chinese memory chips are quietly eating global market share and nobody's talking about it enough.
JPM just dropped a bomb in their memory report. $CXMT (DRAM) and YMTC (NAND) are scaling FAST.
By 2028: $CXMT hits 16% global DRAM capacity, 11% bit supply YMTC hits 16% global NAND capacity AND bit supply
Yes, $CXMT is still 2-3 years behind on bleeding-edge stuff like high-density server DRAM and HBM3E+. But YMTC? Their bit/wafer is basically on par with Samsung and Micron. The tech gap is closing.
Here's where it gets spicy: they're NOT just selling domestically anymore.
HP, Acer, Asus already using $CXMT DRAM in notebooks (non-US markets) Lenovo shipping YMTC NAND in European laptops Apple exploring both suppliers as shortages and pricing squeeze continues
This is a structural shift. Chinese memory isn't just flooding China anymore. It's entering global OEM supply chains. Margin compression for legacy players is coming. Watch how $MU, $WDC, and Samsung react over the next 12-18 months.
If you're long traditional memory names, you need to price in this supply overhang. If you're hunting alpha, start tracking which OEMs are quietly diversifying away from US/Korea suppliers.
That 65-follower gap? That's the entire problem with Web2.
You're told follower count = asset. But if the platform can hide who they are, you don't own an audience. You own a number on someone else's dashboard.
Rented audience in action: • People choose you • Platform controls visibility • Platform controls access • Platform owns the relationship
The most valuable part of a following isn't the count. It's knowing who's there and reaching them without permission from the app.
If 65 followers can vanish into the UI, what else don't you own?
Hidden audience = zero control. This is why we're building onchain social and tokenized communities. Own your graph or get rugged by algos.
You can pay twice and still watch the balance hit zero.
That's not a bug. That's custody.
The platform took your money. The platform shows your balance. The platform can delete it.
People call it a "support issue" because that sounds smaller than admitting the truth.
If your funds only exist because a platform's internal ledger says so, you don't own the funds. You own a claim. Maybe.
This is the same trap creators fall into with revenue. Balances, payouts, tips, subs — all feels real until the system changes, freezes, or ghosts you.
Real ownership starts where vanishing balances stop being normal.
Not your keys, not your coins. Still true in 2025.
Puedes estar listo para vender esta noche y aun así quedar bloqueado por la parte que más importa: cobrar sin exponerte.
Eso no es una molestia con los pagos. Es el modelo de negocio diciéndote quién es el propietario del riel.
Si tus ingresos solo funcionan cuando entregas tu nombre legal, dirección, número de teléfono, sitio web o cualquier otra cosa que un procesador exija, entonces no controlas tus ingresos. Los estás alquilando.
A los creadores se les sigue empujando al mismo mal trato:
Visibilidad a cambio de descubribilidad Cumplimiento a cambio de pagos Identidad a cambio de seguridad
Especialmente en el trabajo para adultos, ese intercambio es absurdo. La persona que crea el valor se ve obligada a asumir el riesgo personal solo para recibir dinero.
No creo que la respuesta sea mejores soluciones alternativas para siempre. Creo que la respuesta es construir rieles para que los creadores puedan cobrar sin ceder la custodia de su identidad.
Hasta que eso exista a gran escala, muchas empresas de creadores no son más que sistemas de permisos con marca.
Si para cobrar necesitas exponerte, no eres dueño del negocio.
Morgan Stanley: Memory correction bottomed. Next leg = buybacks.
July selloff priced in a downturn that never came. Earnings revision breadth peaked in late June while stocks dumped. Memory now trades at ~3x NTM P/E, pricing in an EPS decline that hasn't happened.
Pricing still rising, just decelerating: - 3Q26 DRAM contracts +15% QoQ - NAND +20% QoQ (down from +45-50% surge)
Cycle is late, not over.
Once the market realizes earnings stabilized and prices keep climbing, July fear evaporates. Buyers rotate back in.
Next catalyst: Buybacks.
Every major supplier locked long-term contract volumes. Locked demand + rising FCF = capital returns unlocked.
Samsung, SK Hynix, $MU, SanDisk, Kioxia all disclosed figures on recent earnings calls.
Kurzgesagt necesitaba acceso a un canal de backchannel solo para descubrir que YouTube silenciosamente ha reducido su alcance.
Déjalo asimilar.
No por algún drama de IA. Por la estructura de poder en sí.
Una sola bandera automatizada. Cero aviso claro. Sin una apelación real. Daño a nivel de canal.
Si una plataforma puede eliminar silenciosamente tu alcance y solo las cuentas de primer nivel reciben respuestas, no controlas la distribución. No estás construyendo para tu audiencia. Estás construyendo bajo el martillo de cumplimiento de otra persona.
Por eso, "mejores rev splits" es una forma de consuelo. Un corte más agradable no significa nada cuando una sola decisión opaca de un algoritmo puede separarte de personas que ya decidieron seguirte.
La economía de creadores llama a esto alcance. En realidad, es permiso.
Web3 lo arregla. Tu audiencia es tuya. Tu distribución es tuya. Deja de alquilar la atención a plataformas que pueden desaparecerte de un día para otro.
Las plataformas centralizadas no son tu amiga. Son tu casero.
JPM just raised $SPCX to $240 — and it's all about AI compute eating the world faster than anyone priced in.
The big shift: SpaceX pulled their $1T revenue milestone forward from 2031 to 2030. One year early. Why? Compute is now the primary driver, monetizing harder and faster than the street expected.
JPM bumped 2026 revenue by 14% and 2027 by 64%. Exit-2026 ARR now sits at $100B, up from $75B. Compute capacity hits 5-10 GW by 2027 vs prior 4.2 GW.
Vertical integration + Nvidia partnership = the rocket fuel here. At ~$125 vs new $240 PT, none of this was baked in.
The pain? $300B negative FCF through 2026-2030. Capex peaks at $196B in 2027, $183B in 2028. Positive FCF not until 2031. Front-loaded cycle, high conviction bet.
Risk: everything hinges on Starship. Full reusability unlocks every biz line. Any delay cascades. Add AI power constraints, silicon bottlenecks, Blue Origin + Amazon competition, and Musk's 82% voting control.
High risk, higher upside. JPM sees it. Market doesn't yet.