THE NEXT 18–36 MONTHS

A Book of Trades, and the Theses Behind Them

My raw thoughts on the market, and where I'm putting my own money. Not advice — my book. Author: Aram Barnett · As of: June 14, 2026 · Horizon: 18–36 months (and beyond)

---

"In a world where everyone is levered long the same three narratives, the only edge left is being early, being liquid, and being willing to say the quiet part out loud."

This is not research in the sell-side sense. Nobody is paying me to be nice to a foundation's token, and I hold no allocation from anybody whose feelings I'm obligated to protect. This is my book: what I'm buying, what I'm shorting, what I'm building toward, and the reasoning I'd defend out loud. If you read something here that makes you angry, good. Send me the rebuttal. The whole point of writing it down is to get smart people to tell me where I'm wrong before the market does it for me, with leverage.

A note on method. Every number in here is sourced (citations at the back). Every chart is built from those numbers. Where I'm speculating, I'll tell you I'm speculating. Where I have conviction, I'll size it. Where something is a research note or a company I wish existed, not a live position, I'll say so. And where the honest answer is "I don't know yet," I'll write IDK and move on, because pretending to have a view on everything is how you end up the exit liquidity.

---

THE BOOK, IN ONE BREATH

If you read nothing else, read the table. Everything after it is the homework.

#	Theme	The trade	Direction	Conviction	Horizon
1	AI-driven exploit wave	Long security/audit firms (CertiK, TRM, Chainalysis); underweight unaudited TVL	Long / Short	High	0–18m
2	Agent payments	Long agent-wallet rails (Sponge, Warden); M&A as optionality	Long (private)	High	18–36m
3	Trading agents	Fade the KPI theater; no net-new-volume thesis	Skeptic	Medium	ongoing
4	Prediction markets	Own the operator, never the position	Equity, not trades	High	18–36m
5	Post-quantum / Q-Day	Long PQC as public good; prep an event short on a real break	Research / tail	Medium	tail risk
6	Hyperliquid	High-conviction long HYPE (rotated out of SOL+ETH); target $80–100	Long	High	Q1 2027
7	Ethereum	Mature L1, talent drain to AI; funding source, not a holding	Underweight	Medium	structural
8	Strategy (MSTR)	Short on confirmed sells; buy the BTC dip they create	Pair	High	0–24m
9	Stablecoins	Banks win issuance; CRCL not a buy here	Neutral/Skeptic	Medium	structural
10	XRP	No realizable value thesis	Avoid/Short rallies	Medium	structural
11	CLARITY Act	Re-rating catalyst for compliant venues	Long beneficiaries	Medium	2026
12	DeFi	Boring, battle-tested cash flow; TVL bottoming	Selective long	Medium	structural
13	Memes / pump.fun	Dumb shit sells; treat as a vol sleeve	Tactical long	Low/structural	ongoing
14	Curation (Coinscore)	The market always needs a Moody's	Thesis / build	High	thesis
15	Beyond crypto	Photonics, biocompute, BCI, exploration, longevity, UGC, creators	Personal / venture	varied	36m+

BTC remains the reserve asset of this entire portfolio. None of what follows is a bet against Bitcoin the protocol. Several of these are bets against the people who have wrapped themselves around it.

---

I. AI × CRYPTO

The two largest capital-formation stories of the decade are colliding, and the collision creates three distinct trades. Most people are long "AI + crypto" as a vibe. I want to be long the specific plumbing and short the specific marks. (A note on what's not here: I've dropped the "idle data-center capacity gets repurposed for proof-of-work" thesis I was kicking around. The more I dug into grid economics, the more it fell apart, there's a multi-year backlog for power, and when a data center goes idle it is almost always cheaper to simply switch it off than to run it as a marginal PoW miner. PoW-as-a-buyer-of-last-resort is, in my honest assessment, dead, and power markets are not my circle of competence. I'd rather cut a thesis than defend one I can't underwrite.)

1.1: The AI-driven black-swan wave (the security trade)

I'll be blunt, because the data is blunt. $3.4B was stolen from crypto in 2025, surpassing all of 2024 by the halfway mark, with the Bybit hack alone accounting for $1.5B.² That was before offensive tooling got cheap.

Here is the asymmetry that keeps me up at night and long the security complex at the same time. In February 2026, OpenAI and Paradigm shipped EVMbench, and frontier code models already autonomously exploit >70% of critical Code4rena-class bugs.³ Academic systems (LLM-SmartAudit, SmartLLM) now do multi-agent vulnerability discovery at near-zero marginal cost.⁴ The defensive framing is "AI helps auditors." The honest framing is: a single motivated actor with a frontier model and a wallet can now scan thousands of live contracts, rank them by exploitability, and act, at a cost and speed that did not exist twelve months ago.

This is not hypothetical anymore, and the case study is fresh. On May 29, 2026 a researcher auditing Zcash pointed Anthropic's Claude Opus 4.8 at the code and the model surfaced a four-year-old soundness bug in the Orchard shielded-payment circuit, an under-constrained zero-knowledge proof element that would have allowed unlimited, undetectable counterfeiting of ZEC. It had survived multiple rounds of human auditing. Disclosure on June 5 sent ZEC down ~40% in a day, and the team shipped an emergency hard fork (NU6.2) within days.¹ Sit with what that means: an AI found, in a single audit pass, a protocol-fatal flaw that the best human auditors missed for four years. The supply cap held this time because a defender got there first. The next one may not be a defender. That is the entire thesis in one headline.

I want to be precise about why I'm telling you this, because it's the entire investment thesis: the reason the defensive trade pays is that the offensive capability is now trivially buildable. You don't need a nation-state. You need an API key. When the cost of finding-and-exploiting collapses toward zero and the TVL sitting in unaudited or once-audited contracts stays in the tens of billions, you get a structural transfer from sloppy protocols to whoever is scanning fastest. The market hasn't priced that the attacker's cost curve just fell off a cliff. That mispricing is the trade.

The trades. - Long the scanners and patchers. This is where the revenue actually lands. The incumbent audit and on-chain-intelligence names, CertiK on the smart-contract audit side, and the KYC/AML and forensics players like TRM Labs and Chainalysis, are positioned to turn this threat wave into a durable, recurring revenue line: continuous scanning of deployed contracts, monitoring, and post-incident tracing become must-buy line items for every protocol and every institution touching crypto. Best risk/reward in the AI-crypto basket, in my book. As the attack surface compounds, security spend goes from optional to mandatory, and the scanners sell insurance into a hardening market. - Underweight unaudited / stale-audit TVL. Treat "last audited 2023" as a credit event waiting to happen, and remember Zcash was audited repeatedly. In a portfolio, that means lower exposure to fat-TVL protocols with thin or aging security posture. - Research note: the defensive scanner as a service. Same capability a bad actor would weaponize is a subscription product when pointed at your own (or a client's) contracts with permission. That's the company I'd fund or build, a red-team-as-a-service that finds the bug before the API-key kid does. Not a live position in this book; a gap in the market I keep staring at. To be unambiguous: the value is in disclosure, bug bounties, and patching, not in touching funds that aren't yours. The market for "we found it first, legally" is enormous precisely because the illegal version is now so easy.

1.2: Agent payments: the rails are being poured now

This is the highest-conviction constructive AI-crypto thesis I hold. Agents need to pay, for APIs, for compute, for each other, and eventually on behalf of humans, and they need to do it programmatically, 24/7, in sub-cent increments, across borders. That is a description of a stablecoin on a fast chain. It is not a description of Visa.

The infrastructure went from slideware to production in about a year. Coinbase's x402 revived HTTP 402 for machine micropayments, processed 35M transactions on Solana by March 2026, and was handed to the Linux Foundation in April 2026, a neutral home, which is exactly what a real standard needs.⁵ Stripe and Tempo shipped the Machine Payments Protocol (MPP) for recurring/multi-agent settlement.⁵ Stablecoin settlement hit $33T in 2025 (+72% YoY), with agentic payments named as a growth driver.⁵ On the wallet layer, Warden Protocol raised $4M at a $200M valuation (Jan 2026), processed 60M+ agentic tasks, and onboarded 3M+ users in month one of its agentic wallet.⁶

Why this smells like an acquisition story, and where I'm watching. Stripe already tipped its hand with MPP. The agent-wallet layer, key management, intent execution, policy/limits, cross-chain routing, is precisely the kind of capability a payments incumbent buys rather than builds. The clearest expression of this is Sponge (paysponge.com, YC W2026): two products, Sponge Wallet (each agent gets its own wallet with bank-account access, a card, and crypto across Base/Ethereum and Solana) and Sponge Gateway (businesses list services in a catalog that agents discover and pay for autonomously via x402 and MPP, no code changes). The tell is the team: Jae Choi, Rishab Luthra, and Eric Zhang, are ex-Stripe Crypto engineers, with Zhang the Staff Engineer who personally shipped Stripe's stablecoin payments product and fiat-to-crypto onramp.¹⁵ Read that twice. The team that built Stripe's stablecoin rails left to build the agent-native version. That does not mean Stripe buys them back, founders-left ≠ acquisition thesis on autopilot, but it does mean they know the buyer universe, the product gaps, and the distribution problem from the inside. Plus a launch partner on Solana/Google Cloud's Pay.sh.¹⁵ Underwrite usage first; treat M&A as optionality, not the plan.

The trades. - Long agent-wallet rails with real traction and credible teams, Sponge (ex-Stripe Crypto, cleanest strategic fit in the set) and Warden (on-chain L1 + agentic wallet, 60M+ tasks) are the two names on my desk. Underwrite usage, not airdrop farming. - Size M&A as optionality. The thesis can pay on acquisition or on token/equity appreciation. Venture sizing. Don't need a named buyer to be right about the rails.

1.3: Trading agents: I'm the skeptic at the party

Everyone wants "autonomous trading agents" to be the killer app. I think it's mostly KPI theater. My honest view:

• It will not produce a meaningful amount of net-new genuine on-chain volume. Most "agent volume" will be wash-adjacent activity used to juice foundation KPIs and mislead allocators about real usage. When a foundation tells you transaction counts are up and to the right, ask how much is agents trading with themselves.
• For retail, it's an accelerant on a loss function. If you're a bad trader today, an agent makes you a bad trader faster and at higher frequency. Prediction-market math (next section) applies here too: the sophisticated counterparty eats the unsophisticated one, and the agent just lets retail get eaten 24/7 instead of during waking hours.
• The platform-risk angle is real and it's the trade. When consumer brokerages bolt agents onto retail accounts, the security surface from §1.1 reappears, poorly-sandboxed agent permissions, prompt-injection, over-broad trade authority. The legitimate, defensible position is to be short the platforms that ship under-tested agent features and to be the security firm that gets paid to find those holes before disclosure. The investable expression is underweighting names that are racing agent features to market without the controls, and being long the audit layer that services them.

---

II. PREDICTION MARKETS

I'll save you the suspense: they're stupid, and I'd own the company, never the position.

The data is damning. An SSRN study of every Polymarket trade since 2022 found 68.8% of users lost money while the top 1% captured 77% of all gains.⁷ The cruelest detail: retail actually picked winners more often than the bots, and lost anyway, on price, because they took liquidity with market orders while the sharks provided it with limits.⁷ Over 100,000 wallets lost ≥$1,000; the losers dropped $131M, nearly all of it flowing to a small cluster of high-frequency accounts.⁷ The structure is zero-sum by design and negative-sum after taker fees.⁷ You are not trading against the event. You are trading against Jane Street's cousins, and they have better fills than you.

And yet, the volumes are staggering and real. Prediction markets did >$44B notional in 2025; January 2026 peaked at $26.75B in a single month; Kalshi/Polymarket control ~97.5% of share.⁸ That is a magnificent business, for the house.

There's also a reason I personally don't want the position that has nothing to do with the math: I need liquidity and I need agency. I'm an activist by temperament. I want to be able to act, to push, to change an outcome. A prediction-market position is the opposite, you lock capital against a result you cannot influence. As I put it to a friend: despite my best and most sustained efforts, yelling at the TV has never once changed the outcome of a Wizards game.

The trades. - Own the operator's equity, not the contracts. The toll booth compounds; the gamblers don't. - I built the tooling to prove this to myself: MarketMotion (@marketmotion/cli, github.com/arambarnett/marketmotion-cli), a terminal intelligence platform that searches 25,000+ entities, monitors Polymarket, Kalshi, and Hyperliquid perps in real time, and scans cross-venue mispricings with live alert feeds and webhooks.¹⁶ The conclusion the tool kept handing me: the durable edge is microstructure, execution, and liquidity provision, not opinion. Which is to say it's a quant business, and I'd rather own the quant's landlord than sit at his table. (The flip side: that same infrastructure is exactly what an operator needs, which is why owning the venue, or the data layer feeding the venue, is the trade.)

---

III. POST-QUANTUM & Q-DAY

Let me first say what a quantum computer is, because half the people trading this headline can't.

A classical computer stores information in bits, each a 0 or a 1. A quantum computer uses qubits, which exploit superposition (a qubit can represent a blend of 0 and 1 until measured) and entanglement (qubits' states correlate so the system explores many possibilities at once). For most tasks this buys you nothing. For a specific, narrow class, notably Shor's algorithm, which factors large numbers and computes discrete logs efficiently, it is catastrophic, because the public-key cryptography securing nearly everything (RSA, and the secp256k1 ECDSA that guards your Bitcoin) rests on those problems being hard. Break the math, break the lock.

This is the part people miss: it is not just a crypto problem. It is a threat to the entire installed base of public-key and hash-based security, banking, TLS, signed software, state secrets. Crypto is simply the most honest about it because the targets are sitting on-chain, balances public, waiting.

Now the sober part. Google's Willow (Oct 2025) hit "verifiable quantum advantage" with 105 qubits.⁹ Breaking BTC in a day is estimated to require ~13 million physical qubits, five-plus orders of magnitude away.⁹ But Google itself now puts a ~10% probability on a quantum computer recovering a secp256k1 key from an exposed public key by 2032, and recent papers keep lowering the resource estimates for Shor.⁹ Q-Day is not tomorrow. It is also not "never," and the market prices it at roughly never.

What changed my timeline is not Google, it's the roadmaps from the two players with the deepest pockets. IBM laid out a credible path to a large-scale fault-tolerant machine, "Starling," by 2029, using qLDPC error-correcting codes that cut physical-qubit overhead by up to ~90%; its near-term Nighthawk processors push toward genuine quantum advantage, and the modular stack (Loon, Kookaburra, Cockatoo) is shipping on schedule.¹⁸ Microsoft went a different route with topological qubits: Majorana 1 (Feb 2025) was the first chip on a topological core, and Majorana 2 (June 2026) claims ~1,000x better qubit stability, with Microsoft now targeting a scalable machine as early as 2029, halving its previous timeline.¹⁹ Two independent, well-funded roadmaps both converging on 2029 is the signal. The fault-tolerance overhead, the thing standing between 105 physical qubits and millions of useful ones, is the variable everyone is racing to collapse, and the trend is one-directional.

Why I won't chase a venture return here. Post-quantum cryptography is a public good. The whole point is that the standards become ubiquitous, open, and free, NIST-style. A good that should be widely and freely distributed is a bad place to look for a clean liquidity event. I'll fund and support open-source PQC because it's correct, not because I expect a markup.

Where the actual money might be: the panic, not the cure. The investable asymmetry isn't PQC startups, it's the tape on the day someone demonstrates a real break. The market will not wait for the orderly migration to a quantum-resistant ledger. It will sell first. The large-cap tokens most exposed to "your keys are now guessable" will gap down, and the contagion won't stop at crypto, anything perceived as cryptographically naked (and a lot of tech that isn't but will be sold anyway) gets hit. Timing is the whole problem. This is a tail, not a calendar trade.

The book / the prep. - Long open-source PQC as a public good, eyes open, no liquidity event expected. - Research note: build the listener. The source of truth is the research frontier. I'd run a scanner over arXiv / lab publications / Chinese and US national-lab output that flags the qubit-count and algorithmic-improvement milestones that actually move the estimate. (Research-monitoring system, not a position.) - Have the short structure sketched, trigger on a credible-break signal, don't sit short waiting for Godot. Most-exposed large caps and the obvious tech adjacencies. The migration to quantum-resistant ledgers is bullish long-term for the chains that execute it, but the news of the break is a violent down-leg first. Buy that dip in BTC specifically; the protocol can and will harden. The drop is temporary; my BTC position is not.

---

IV. HYPERLIQUID: THE TICKET

I rotated my Solana and Ethereum into HYPE. That's not a trim. That's a conviction rotation, and I'll defend it. It is also a concentrated, high-beta bet, not a diversified savings account. If crypto has a risk-off week, this thing eats first. I know that going in.

The numbers as of mid-June 2026: HYPE ~$60, market cap ~$15.3B, fresh ATH of $75.43 on June 2, 2026, with a $4.4B USDC treasury and a buyback program directing up to 90% of reserve yield into HYPE buybacks starting October 2026, scaling with TVL.¹⁰ That last point matters more than the price: it's a reflexive flywheel, more volume → more reserves → more yield → more buybacks → fewer tokens. That is a real business returning real cash flow to the token, which is more than most of this asset class can say.

The thesis in plain English: - Mature, focused team. They build for the trader, not for a conference keynote. The product is fast, the UX is serious, the roadmap is disciplined. - Derivatives mindshare, on-chain. The last cycle proved how much flow consolidates around one venue that traders actually want to use. Hyperliquid is capturing that on-chain perps mindshare self-custodial and transparent, without the custodian-blows-up failure mode. Underwrite the product and the flow, not a personality cult. - Cash flows, not narratives. The buyback ties token value to actual platform economics. Underwrite it like an exchange, because it is one.

The trade. Long HYPE, target $80–100 by Q1 2027. That's roughly a 33–66% move from ~$60, underwritten by the buyback flywheel and continued share-take in on-chain perps. I'm not pretending that's a moonshot multiple; it's a cash-flow and share-take underwrite on a name I already sized as core. Risk: high-beta exchange token; broad crypto drawdown hits it hard first. Concentration risk is real because I funded it from SOL+ETH. I respect the volatility and I am not adding with money I can't watch go red.

---

V. ETHEREUM: RESPECTFULLY, IT'S OLD

I don't hate Ethereum. I'm just not paid to be nostalgic. ETH printed back under $2,000 in June 2026 (~$1,664 on the 14th), having opened the month above $2,000, as capital rotated toward AI and higher-beta alternatives.¹¹

Steelman first, because I'm not dumb: ETH still settles a huge share of DeFi and stablecoin activity, the ETF bid is real and reflexive, L2s extend the franchise even as they cannibalize L1 fees, and "underweight ETH" is a crowded take that can get you run over on any risk-on day. If your mandate is "own the settlement layer of crypto," ETH is still the default answer for a lot of serious money.

Why I'm still underweight in my book anyway:

1. The easy reflexive growth is behind it. The monetary-premium story is mature, the fee story is cannibalized by L2s it subsidized into existence, and the asset trades like infrastructure, not like an option. Infrastructure can be valuable and still be a mediocre risk/reward from here.
2. The talent moved on. The most ambitious builders I know rotated to AI. Crypto-native genius is now being spent on agents, inference, and compute markets, not on yet another rollup. Chains are ultimately talent markets with tokens attached, and ETH is losing the talent war to AI, full stop.
3. I'd rather own the forward cash-flow L1 expression. In my book that's HYPE's exchange economics, not ETH's mature settlement premium.

The trade. Underweight, structurally. I'm not pressing a dedicated short, crowded, dangerous, ETF bid underneath it, but ETH is a funding source in my book, not a holding. I sold mine into HYPE. If you must own L1 beta, own the one where the team and the cash flows are pointed forward.

---

VI. SAYLOR & STRATEGY (MSTR): THE FORCED SELLER FORMING

Let me be careful and precise, because this is the most actionable short in the book and I don't want it dismissed as a personality dispute. Bitcoin the protocol is unchanged and I love it. Strategy the levered equity is a different instrument, and I think its steward has become a net risk to the price.

The premium that powered the whole flywheel is gone. mNAV compressed from 3.4x (Jan 2024) to 0.78x (June 2026), Strategy now trades at a discount to the Bitcoin it holds.¹² The model only works above 1.0x: issue equity/converts at a premium, buy BTC, raise BTC-per-share, repeat. Below 1.0x, the machine runs in reverse, you can't accretively issue, and you still owe the coupon. Against 845,256 BTC sits $8.2B in convertible debt and $13.5B in preferred demanding dividends, and in late May 2026 Strategy sold BTC for the first time since 2022 to service preferred payments.¹² The "never sell" covenant, the entire brand, broke.

Perception is reality. It does not matter whether 32 BTC is "material" to the float. When the most-watched corporate holder on earth becomes a seller, that is a signal to every leveraged long in the complex. Each confirmed sale carries a reflexive price impact far beyond its notional, because it rewrites the narrative from "infinite demand sink" to "forced seller servicing debt." And debt service is not a one-time event, it's a calendar. As long as the discount persists and the coupons come due, the probability of further liquidation rises with time.

The trades. - Monitor the addresses and the filings. Strategy's holdings and movements are public, and so are its 8-Ks. I'd run an address-and-filings listener that flags outbound movement and any new sale disclosure. (Pure public-data monitoring.) - Short MSTR / the complex on confirmed sells. Open shorts on disclosed liquidation, sized to the reflexivity, not the notional. - Buy the BTC dip they create. This is the crucial pair. The forced selling produces a temporary BTC drawdown driven by an equity-structure problem, not a protocol problem. I use that dip to add to spot BTC. Short the levered wrapper; accumulate the underlying. Nothing changes at the protocol level, only the cap table does.

---

VII. STABLECOINS: IDK, bruh

I'll quote myself, because it's the honest version: "IDK, bruh." I was bullish on Circle; since the IPO it's been a grind.

CRCL IPO'd at $31 (June 2025), closed day one at $82.84 (+167%), ran to an ATH of $298.99, then round-tripped to ~$62 by Feb 2026 before a partial recovery to ~$104 by late May, a ~79% drawdown from the high.¹³ The chart is a near-perfect "the IPO was the top" pattern.

The structural worry is bigger than one stock: stablecoin issuance is likely to be dominated by existing banks. Once a regulated framework lands, JPMorgan, Citi, and the rest have the balance sheets, the distribution, the float economics, and the regulatory cover to issue. A standalone issuer's moat, float income on reserves, is exactly what a bank is structurally better positioned to capture. In that world Circle is a margin-compressed toll operator competing with institutions that print their own demand.

The trade. Neutral-to-skeptical on pure-play issuers at these levels. The cleaner expression of "stablecoins win" is the distribution and settlement layer (see agent payments, §1.2) and the banks that issue, not the independent issuer fighting them on price. I'm not short CRCL with conviction, I just don't see the asymmetric long here.

---

VIII. XRP: CAN WE ALL JUST DUMP THAT SHIT?

I'll keep this one short and let the zealots fill my mentions. XRP holders are the most relentless community in this asset class, and I respect the conviction the way I respect a hurricane. But conviction is not cash flow. The value of the token has yet to be realized and, in my view, never will in a way that justifies the market cap, it remains a token in search of the utility its holders have insisted is imminent for a decade.

Meanwhile leadership reads as checked-out: the CEO seems more interested in his luxury-car collection, and the CTO appears to be speed-running an influencer arc. Tongue half in cheek: if there were a liquid market in clout, I'd be long it to the moon, but clout is not on-chain value, and you can't service holders with vibes.

The trade. Avoid; fade rallies that run on litigation-headline and ETF-hopium rather than usage. I'm not building a dedicated short, the community can squeeze you on pure faith, but it has no place in a fundamentals book.

---

IX. THE CLARITY ACT: THE RE-RATING CATALYST

What it is. The Digital Asset Market Clarity Act is the U.S. market-structure bill that finally divides jurisdiction over digital assets between the SEC and CFTC and defines when a token is a security vs. a commodity. The House passed it in July 2025; the Senate Banking Committee advanced it 15–9 on May 14, 2026, and it's now being reconciled with the Senate Ag Committee's companion bill before a full-Senate vote and House reconciliation. As of June 2026 it is not yet law, with lawmakers targeting a summer 2026 finish.¹⁴

How it impacts us. Regulatory ambiguity has been the single largest discount applied to U.S.-touching crypto businesses. Clarity collapses that discount. It (a) gives compliant exchanges and custodians a defensible operating perimeter, (b) lets institutions allocate without career-risk on "is this an unregistered security," and (c) separates the legitimate venues from the gray-market ones. It is, functionally, a re-rating catalyst for anyone who's been operating cleanly and waiting for permission.

Where to invest. Long the beneficiaries of legitimacy: compliant U.S. exchanges/custodians, the tokenization and settlement rails that need legal clarity to scale, and the curation/compliance layer (§XII). The asymmetry is in names that have eaten the compliance cost already and re-rate the day the discount lifts. Risk: the bill is hostage to a crowded Senate calendar and could slip past the August recess, so treat it as catalyst optionality, not a date you underwrite. (I'll add a vote-tracker so we're positioned ahead of the print.)

---

X. DEFI: BORING IS THE BULL CASE

DeFi is boring now. That's the entire bull case. The mercenary yield farmers left for memecoins and agents; what remains is the protocols that survived multiple cycles, drawdowns, and exploit waves and just kept matching lenders to borrowers and traders to liquidity. Boring means battle-tested. Boring means the contracts have been adversarially probed for years (though see §1.1, nothing is permanently safe). Boring means fees without drama.

Where we are in the cycle: at or near the bottom. The numbers look ugly, and that's the opportunity. DeFi TVL fell to roughly $72.5B in Q2 2026, down from ~$178B, gutted by a brutal run of exploits, the KelpDAO hack alone triggered a ~$13.2B two-day unwind, with Aave shedding $8.45B in deposits in 48 hours.²⁰ But look underneath the panic: stablecoin supply is holding near $315B, with $150B+ in Treasuries backing USDT/USDC and on-chain volumes still growing 20–30% a month.²⁰ That is the tell. The liquidity did not leave the building, it moved to the sidelines and de-risked. When that much dry powder sits next to a halved TVL, you are looking at a market that has flushed leverage, not one that has lost users. My read: we are at or near the bottom, TVL levels out here, and the next leg is up as the survivors consolidate share and yield-seeking capital rotates back into the names that didn't break.

Where I'm positioned: Superform. I'm an investor in Superform (a cross-chain yield/wealth-management protocol now operating as a user-owned, self-custodial on-chain neobank). Even through the worst DeFi drawdown in years, Superform carries roughly $10B in TVL across 50+ protocols and 800+ earning opportunities, with 180,000+ depositors, and expanded to the U.S. with a mobile app.²¹ I'll be honest about the tape: sector-wide growth has been negative, and a universal aggregator faces real competition from curated vaults. But that is exactly why the survivors matter, holding ~$10B of self-custodied deposits while the rest of DeFi bled is not nothing, it's a demonstration that the product retains capital when capital is most scared. In a bottoming market, you want the aggregator that kept its TVL, because it is the one positioned to compound when the flows come back.

The trade. Own the blue-chip, cash-flowing DeFi that converts fees to holders and has a defensible security posture (Superform is my expression of it); underweight anything reinventing a primitive that already works. In a portfolio sense, DeFi is the ballast, low narrative beta, real usage, bought near a washed-out bottom. Pair it with the security overlay from §1.1 and you have a position that gets safer as the attack surface gets more dangerous, because the survivors consolidate share.

---

XI. MEMES & PUMP.FUN: LONG LIVE THE CASINO

Long live memes. Dumb shit always sells. I'm not going to intellectualize this into something it isn't. Memecoins are a permanent, structural feature of crypto because they are the purest expression of what this technology actually does well: spin up a liquid, global, 24/7 market on any narrative in seconds, with no gatekeeper. That's also the problem (see §XII), but as a trade it's an evergreen volatility sleeve.

The trade. Treat memes as a small, disciplined, rules-based vol allocation, not a thesis, a sleeve. The platform-level observation is the more interesting one: the launchpads (pump.fun-style venues) are toll booths on degeneracy, and like prediction markets (§II), I'd rather own the casino than play the table. The house captures fees on every coin whether it pumps or rugs.

---

XII. INDUSTRY ISSUES: AND WHY CURATION IS THE BUSINESS

Four things are structurally broken in this industry, and each is an opportunity for someone willing to fix it:

1. Talent is thin. Too many tourists, too few builders, and the best ones just left for AI (§V).
2. Fragmentation. Liquidity, standards, and attention are splintered across dozens of chains, none with decisive share. Capital efficiency suffers; so does the user.
3. Grifters. The asset class still over-indexes on people who are very good at raising and very bad at building.
4. The gatekeeping paradox. It is trivial to deploy a token, and so a lot of these things are garbage. The absence of any quality filter is freedom, and it is also why retail gets fleeced and why institutions stay out.

That last point is the whole thesis, and I've been early to it before. Ten years ago I started a company called Coinscore to do exactly this: ratings, assessments, and due diligence for crypto assets. The market wasn't ready and the tooling wasn't there. Both have changed. With Clarity (§IX) creating a compliance perimeter, with institutions arriving, and with AI making real-time assessment of code, treasuries, and on-chain behavior actually tractable, crypto needs a Moody's/Morningstar more than ever. Curation is not a feature; in a world of infinite, frictionless token issuance, curation is the scarce good.

The trade. Own / build toward the curation layer. A credible, AI-native rating and due-diligence business sits at the intersection of every other thesis in this memo, it consumes the security signal (§1.1), the regulatory signal (§IX), and the fundamentals work, and sells trust into a market drowning in noise. This is the one where I'd rather be the operator than the allocator. Not a fundraising ask in this piece; a thesis I've held for a decade and still think is right.

---

XIII. BEYOND CRYPTO: THE ASYMMETRIC BOOK

Crypto is where I have an edge today. These are where I think the next decade of asymmetry lives, and where I'm already putting (or want to put) personal capital. Shorter treatment, same honesty: some of this is underwritten, some of it is conviction and curiosity, and I'm not going to sand my personality off the page to make it look like a bank research note.

• Photonics, the inference bottleneck is physics, and physics has a fix. I'm an early investor in Volantis (volantissemi.ai). The thesis is brutally simple: AI inference is increasingly gated not by FLOPs but by chip-to-chip communication, moving data between dies over copper is hot, slow, and power-hungry. Volantis builds a photonic motherboard that moves data between chips with light instead of electrical signals, "inference at the speed of light." Founder Tapa Ghosh comes in with 15 prior idea-to-silicon photonic tapeouts, and the cap table tells you who else sees it: backers include Sam Altman and Alex Wang (Scale AI).¹⁷ If the bottleneck to AGI-scale inference is interconnect, and it increasingly is, the company selling the interconnect at the speed of light is selling the scarcest pick in the entire AI buildout.
• Biocomputing & brain–computer interfaces. Substrates beyond silicon, and direct neural I/O. Early, capital-intensive, binary outcomes, venture sizing only, but the TAM is "all of computing" and "all of human–machine interface."
• Telepathy, precognition, anticipating the future. Yeah, I said it. The spicy one. I'd back the serious end of this, DARPA-grade research on anticipatory cognition brought into the public domain and applied to society. I'm a fan of Ryan Rossner's work here. Frame: research-stage, lottery-ticket sizing, asymmetric payoff. If that makes you close the tab, we were never going to allocate the same way anyway.
• Terrestrial exploration, "as above, so below." Everyone points telescopes up; I think the boom is down and out, satellites that scan the Earth, heavy earth-movers, oceanography, personal submarines. I want to back teams of ex-military, ex-big-oil, and geologists starting companies here, and to fund expeditions: American cave systems, the Amazon and Congo, Caribbean nautical work. (The expeditions also make extraordinary livestream/IP, a media flywheel on top of the science.)
• Farming. Part passion, part sovereignty hedge, cattle, sheep, chicken. Self-reliance is a real option value when the rest of the book is levered to frontier tech. Not a Sharpe-ratio trade. A life trade that also hedges.
• Experience hotels. A hospitality brand built around experiences, starting with a Napa/Sonoma vineyard. Real assets, cash yield, brand optionality. Looking for the right partners, not pitching a round in this memo.
• Chinese EV infrastructure in America. Charging/servicing/parts infrastructure positioned for the cost curve, navigating the obvious policy risk, which is exactly why it's mispriced.
• Longevity & peptides, living past 100. The science is finally crossing from supplement-grift into real intervention. Selective, evidence-led exposure.
• GTA VI mods / UGC studios. People are minting money on Roblox and Fortnite UGC; GTA VI will mint the next cohort. I want to invest in or acquire UGC studios producing now and primed for the GTA VI moment, buy the talent and pipeline before the platform opens.
• Creator economy, Curbily (where I'm both investor and operator/CTO). The one I'm closest to, so the numbers, not the hype. Curbily is the operating system for branded content, the agentic layer where brands, creators, and production meet end to end. A brand writes a campaign brief and instantly gets ranked, confidence-scored creator matches (the Curbily Score, the §XII curation thesis again), reaches out through a guided flow, closes in one inbox, and produces with Studio ("Hollywood in a box"). Discovery, deal, production, live events, one system run by AI agents.
  • Why now: brands are becoming creators, building in-house studios, which explodes demand for discovery, deal-making, and production tooling stuck in spreadsheets, DMs, and agencies. The market is ready for an operating layer, and for AI agents to run it.
  • What's live (not roadmap): six AI agents already working end to end, brands@ (brief to matches to auto-proposal), managers@ (any-format roster to claimable creators), productions@ (script/event to full production package), castandcrew@ (conversational talent onboarding over email and SMS), hello@ (smart router), and a Creator Inbox Agent that reads a creator's inbound brand deals, scores them, and negotiates rate, deliverables, and schedule over SMS against the creator's own rules, with kill-switch, caps, and full audit governance. The "@curbily.com agentic front office" is shipped and in pilot.
  • Traction: ~$50M in content/script budgets defined on the platform in a single month (top-of-funnel scope, not spend); ~2,900 creators across ~60 agency rosters, enriched with media kits and live social stats; early monetization validated via per-campaign match packages (~97% margin), now deliberately moving to the bigger prize, a take-rate on the transaction itself, with Stripe payment rails live and Deel contracts/escrow integrating. Team has shipped with Hulu, Netflix, Tubi, and DreamWorks.
  • Why it belongs in this book: owning every creator's deal inbox = owning the transaction layer of the creator economy. That agentic front office is the same machine-to-machine commerce thesis from §1.2, pointed at a $10B+ vertical, and it's already running. (Newest expansion: full live-event production, talent, vendors, permits, budget, and an exec pitch deck from a single brief.) This is where a chunk of my time and capital actually sits. If you want the company story separately, email me — this piece is the market book, not a pitch deck.

---

LET'S TALK

The thesis underneath all of it is the same one I've held for a decade: crypto, and now frontier tech broadly, generates infinite noise and scarce curation. The edge is being early, staying liquid, having the stomach to short your friends' bags, and the discipline to write down why before the market tells you. If any of this resonates, or if you think I'm dead wrong, I want the conversation.

Reach me at aram.barnett@gmail.com.

---

A WORD ON RISK (because the lawyers, and reality, demand it)

These are my personal opinions and my personal book, written to share how I'm thinking and allocating. This is not investment advice, an offer, or a solicitation. I hold or may hold positions in assets discussed (notably HYPE and BTC, equity/venture interests including Superform, Volantis, and Curbily, and short interest in MSTR-complex names on the conditions described). Crypto assets are volatile and can go to zero. Every price, target, and probability here is a view, not a promise, and several depend on events (a Clarity vote, a quantum demonstration, a forced sale) that may not occur on the timeline I expect, or at all. Do your own work. Then tell me where I'm wrong.

---

SOURCES & CITATIONS

---

Built with market data as of June 14, 2026. Charts are the author's, from the cited figures. Private-company figures (Curbily, Volantis, Sponge, MarketMotion, Superform) are drawn from the author's own materials and public sources.