Sequoia: "The Next $1T Company Sells Work, Not Software"
Sequoia Capital — the firm that backed Apple, Google, Nvidia, YouTube, Airbnb, Stripe — dropped a thesis worth reading closely. The old model: For 20+ years, tech sold software. Microsoft sells Office, Adobe sells Photoshop, Salesforce sells CRM. Tools that help humans work faster. Copilot for everything. The problem: Customers don't want software. They want work done. You don't want accounting software — you want books closed on time, taxes filed, reports delivered. The insight: For every $1 businesses spend on software, they spend $6 on services. SaaS has been fighting over that $1. AI can now digitize the $6 — the knowledge workforce itself. The map: Sequoia's Opportunity Map plots work by Intelligence vs. Judgement, Outsourced vs. Insourced. The sweet spot: highly standardized, already outsourced workflows — Insurance Brokerage ($140-200B), Accounting ($50-80B), Healthcare Revenue Cycle ($50-80B). 2025 = Copilot. 2026 = Autopilot. The winners won't build AI tools for accountants — they'll build AI accounting firms. Sequoia warns most Copilot startups face the Innovator's Dilemma: today you sell software to accountants, tomorrow you'd compete with them. The bottom line: The next $1T AI company won't have the smartest chatbot. It'll be the first to turn work into a service you buy like electricity. Crypto Cashtags That Align Tier 1 — Direct "Sell Work" AI Agent Plays: FET — Fetch.ai (ASI Alliance). Autonomous agents automating enterprise workflows. Built for agents doing work, not providing tools. (Ethereum)$VIRTUAL — Virtuals Protocol on Base. The agent creation infrastructure — a factory for building AI agents that sell outcomes, not subscriptions. (Base)$GRIFFAIN — Griffain on Solana. AI agent that executes on-chain actions. "Sell work" in its purest crypto form — you describe what you want, the agent does the job. SolanaOLAS — Autonolas on Ethereum. A framework for coordinating autonomous agent fleets — think of it as the operating system for a decentralized AI workforce. (Ethereum)Tier 2 — AI Infrastructure (the picks & shovels): $TAO — Bittensor. A decentralized AI network where agents train, compete, and earn. The network layer for autonomous work — the protocol that lets AI sell its output peer-to-peer.RENDER — Render Network on Solana. Decentralized GPU compute. Every agent running in production needs compute power — Render is the hardware layer. (Solana)Tier 3 — AI Agent Ecosystem: $AI16Z — ai16z/ElizaOS on Solana. Named after the VC model itself — a DAO-run AI agent fund that makes autonomous investment decisions. Pure meta-commentary on Sequoia's thesis. Solana$ZEREBRO — Zerebro on Solana. An autonomous AI agent creating and distributing content without human intervention. "Sell work" in its most literal form — the agent is the output. Solana The Sequoia filter: The next $1T company sells work, not software. In crypto, that means looking past the tool tokens and toward the agent workforce tokens — the protocols where AI doesn't just assist, but replaces the $6 of services for every $1 of software. Not financial advice. #NewsAboutCrypto #StrategicInvesting #BTC #SequoiaCapital
Kevin O'Leary's Perspective on the Semiconductor Market and Global AI Dominance
Kevin O’Leary’s Take on $NVDA , $AMD , & $INTC
💥The "Queen Bee" Strategy: O'Leary views chips (NVDA, AMD, INTC) as "Queen Bees." Their real value is the "Honeybees"—the global developer community that writes software for them.
💥The Policy Error: He argues that banning high-end chip sales to adversaries is "bad policy". If the U.S. removes its "Queen Bee," competitors like Huawei will fill the void, causing developers to leave the American ecosystem.
💥AI Dominance: Currently, Nvidia chips drive the vast majority of AI productivity. To win the future of warfare (drones and robots), O'Leary believes the U.S. must keep global talent tied to American hardware to maintain its "Hive" as the global standard.
The 30-year U.S. Treasury yield hit 5.31% on Monday — the highest since mid-2007 , inching toward the 5.44% peak of the GFC era. This isn't a blip; it's a regime statement.
The numbers: 📉$25B 30Y auction last week priced at 5.216% — richest since 2001 ; the 10Y auction (highest since 2007) followed days later
💸National debt racing toward $40T , annual interest bill already past $1T
🏭$145B record August IG corporate issuance — AI capex debt crowding out the long end
🌍Global echo: Japan's 30Y at record 4.05% , Canada's highest since 2010
Why it matters: The Fed has cut 175bp from the peak — and the long end still climbs. That's the market pricing fiscal supply + term premium + sticky inflation , not Fed policy. Equities took the hit Monday (Dow -0.5%, Nasdaq -0.3%), tech/duration names most exposed, while mortgage rates and the $90+ Brent backdrop keep the feedback loop alive.
Watch level: 30Y holding >5.3% = every long-duration asset reprices. Breaking toward 5.44% = 2007 redux vibes. Barclays' rates desk is blunt: "We have been arguing against fading the long-end sell-off."
The Fed controls the short end. The market controls the long end. Right now, the market is winning. 🔔
The Ethereum Foundation just switched on Platåberget — a brand-new public testnet purpose-built to shake down Glamsterdam , the L1-scaling hard fork slated for mainnet in H2 2026. Hard fork activates on the testnet August 20 .
What's inside Glamsterdam: 💥ePBS — enshrined proposer-builder separation: block building gets trust-minimized at the protocol level, killing the MEV-boost middleman 💥Block-level Access Lists (BALs) — leaner transaction execution paths 💥Gas repricing toward ~200M gas + contract deploy size cap up from 24KiB → 64KiB (initcode 48KiB → 128KiB) 💥EIP-8037 — a new state-gas dimension: writing fresh state costs real bytes, not just gas 💥All wrapped in meta EIP-7773 — the biggest protocol surface change since Fusaka
Why it matters: 🧪Safety-first sequencing: Platåberget runs "several months" → then Sepolia & Hoodi → then mainnet. Bugs get killed in a sandbox, not in production 🔓Open to everyone: small validator set, public participation, one-click RPC/faucet/client images — community feedback via Ethereum R&D Discord 🚀The scaling thesis: gas repricing + bigger contracts + ePBS = Ethereum preparing for the next era of L1 demand (the 2027 "Hegotá" upgrade, with 66 EIPs already in the queue, is already waiting in the wings)
History note: post-Pectra, $ETH rallied hard off the last big fork cycle. Upgrades don't pump on their own — but they set the stage. This testnet is the first real proof Glamsterdam ships on time.
Not every launch needs fireworks. Sometimes it's a quiet testnet in the mountains — where the next era of Ethereum gets its first stress test. 🏔️
$SAMSUNG : The oldest rule in the book. What goes up must correct. What gets sold off must bounce back. Samsung just ripped +11% in five sessions off a 28% drawdown from July's peak — but this is a ₩1,700T+ gorilla, not a meme. Big caps don't trend one-way; they mean-revert, violently, and fast.
Trigger: reclaim of ₩274,500 (Aug 14 close) → momentum confirms Long bias: ₩262,000 – ₩271,000 zone (pullback buy) Invalidation: daily close < ₩246,000 (Aug 5 low) — thesis dead
$CYS : The sellers are out of ammo. Selling pressure is fading, buying pressure is starting to build — and the chips that dumped this thing are nearly fully absorbed. Expect a 20–30% recovery from here. Long now, before the bounce turns into a breakout.
A secret US-brokered deal just defused a ticking nuclear leftover: the IAEA will pull yellowcake uranium out of Syria's clandestine "Site 99" — remnants of Assad's Al-Kibar reactor, destroyed by Israel back in 2007.
The backstory: Israel already cratered the site entrances after Assad fell. When Israeli intel flagged new activity, Washington chose diplomacy over another airstrike — and Damascus didn't even know the material was there. Deal signed ~3 weeks ago; removal targeted before end of 2026.
Why it matters: 🛢️Material can't fuel a warhead, but it's dirty-bomb capable — real radiological risk off the table 🤝New Syria gets a path out of the old regime's shadow — legitimacy over chaos 🎯The playbook Israeli officials now want to copy for Iran's HEU , if Tehran ever takes a deal — that's the real de-escalation signal for the region (and for oil)
Nuclear forensics on the yellowcake could also answer where it originated — potentially reopening the Uranium One question.
Not every headline ends with bombs. Sometimes it ends with a forklift and an IAEA inspection tag.
$CL : Pure Headline Tape. Trump just said "No" to extending the 60-day US–Iran truce — oil spiked overnight. But when the entire rally is one headline, it unwinds just as fast. News in = fade on. The candle is priced; the echo is not. Short the exhaustion, not the narrative.
#twodroneshitkurdistanpmoffice Oil ($BZ ) just ripped $87.3 → $89.2 in one 1H candle on the Kurdistan/Iran headlines — and the pullback is holding textbook structure. The question isn't if $90 gets tested, it's how it gets there.
💥Entry zone $89.05 – $89.25 (breakout retest) 💥Stop loss $88.75 (below $88.89 swing low) 💥TP1 $89.55 (re-test of spike high) 💥TP2 $90.00 (psychological + headline magnet) 💥TP3 $90.60 (extension)
Two drones just hit the office of Kurdistan Region PM Masrour Barzani in Erbil's Pirmam district. Kurdish officials say the strike came from Iranian territory. Tehran's answer: "highly suspicious" — a "false flag."
Iraq's PM Ali al-Zaidi has ordered a full investigation into who launched the drones, with findings to be made public. Same day: an alleged militia assassination attempt on Iraq's top judge, plus Kataib Hezbollah threats against the PM himself. The balancing act between Washington and Tehran is unraveling in real time.
For markets, the script is unchanged: every time this "cold war" turns hot, crude catches a bid. $USO.ETF is the direct hedge on a Persian Gulf risk premium that refuses to die — oil was already climbing this week on Hormuz impasse and Houthi strikes on Aramco infrastructure.
The key risk to watch now: if the Iraqi probe publicly points at Iran, Baghdad–Tehran relations go from strained to openly hostile. That's the kind of headline that reprices barrels.
ColdCard was supposed to be the most trust-minimized wallet in Bitcoin. Right now it's the most expensive bug in hardware-wallet history.
Galaxy Research just confirmed losses have blown past $115M — and warns the attack is "still ongoing." Their high-confidence minimum sits at 1,730 BTC, with 1,531 BTC still sitting in attacker-controlled addresses . Estimated total: up to 1,816 BTC across 5,200+ addresses.
The scar tissue is already visible: ~ 233,000 $BTC (~$15B) has moved off long-term wallets since the exploit — multisig migration, exchange withdrawals, anything but ColdCard.
The vulnerability lived in firmware 4.1–4.1.9 (Mar 2021 – Jul 2026) for years. If you generated a seed phrase on those versions: treat it as compromised and migrate. Today.
The motto was "don't trust, verify." This is what verifying costs.
The dollar just hit its lowest since May — its 6th red week in 7 — and the market is misreading why.
This is not a dovish-Fed story. Yes, July retail sales fell 0.6%, the worst in over a year, and September hike odds collapsed from ~50% to under 30% in a single week. But look closer: 30-year Treasury yields are at 2007 highs while the dollar craters. That divergence is the real signal.
A falling dollar with rising long yields isn't the "Fed saves us" trade — it's the fiscal + political trade: soft data, deficit anxiety, and Trump's pressure campaign on the Fed (regular Warsh calls, renewed attempts to fire Governor Cook) all eroding confidence in the currency itself. When the dollar dies and long rates refuse to fall, markets are pricing one thing: debasement .
Gold at ~$4,400+ and pressing record highs is the clean tell. The trade of this macro regime is simple — short the dollar, own real assets .
🚀🚀The on-chain way to play it — $PAXG (1H)
💥Long entries $4,407–$4,417 💥Stop at $4,390 (below the $4,393 swing low). 💥Targets $4,430 → $4,445 → $4,460. 💥Confirmation trigger: 1H close above $4,423 extends the breakout 💥R:R from a ~$4,410 entry is roughly 1:1 at TP1 up to 1:2.5 at TP3, with a tight 0.45% stop — clean for a 1H momentum trade.
The Ethereum ($ETH ) Foundation just opened Platåberget , the first public testnet for the Glamsterdam upgrade, with the hard fork set to go live on August 20. This is the biggest L1 tune-up since The Merge, and it's not about meme coins or hype — it's about giving Ethereum more execution capacity while stripping power from MEV cartels.
Three things actually matter here. First, ePBS (enshrined proposer-builder separation) removes the builder middlemen that have been extracting value and centralizing block production — validators keep more, cartels lose their grip. Second, gas repricing toward a ~200M gas floor roughly triples the blockspace per block, which means cheaper and more efficient L2s on top. Third, contract size jumps from 24KiB to 64KiB , so developers can ship far more complex on-chain logic. For devs, there's a catch: hardcoded gas limits are about to break, and meta EIP-7773 is the heads-up for wallets, indexers, and gas estimators.
History has a pattern: every time Ethereum ships a major capacity upgrade, the L1-native names re-rate for weeks. This fork is exactly that kind of catalyst — and it's still underpriced.
The LONG 1H setup — $LINK 💥Entries at $9.47–$9.52 💥Stop at $9.40 below the Aug 17 low. 💥Targets are $9.62, then $9.72 (the Aug 15 high), then $9.85.