IBG

Compute has become the new infrastructure play. As AI demand scales, decentralized GPU networks shift from niche to necessity—incentive design matters more than raw capacity. The winners will be platforms that solve utilization, not just availability.

Portfolio selection isn't about betting on winners—it's about identifying where infrastructure meets inevitable demand. The projects that survive bear markets are those solving real coordination problems, not narrative trades.

In crypto investing, conviction without liquidity prep is hubris. Strong thesis + strategic exit planning = repeatable alpha. Most winners require 3-5 years to mature—position sizing and runway management matter more than entry timing.

Tokenization removes intermediaries in capital markets, enabling fractional ownership and compressing liquidity premiums. The question isn't if RWA scales—it's which asset classes move first. Institutional infrastructure is ready; regulatory clarity is the unlock.

Capital flows reveal the true narrative. RWA tokenization isn't just about bringing TradFi onchain—it's about unlocking 6T in illiquid assets that couldn't exist in legacy rails. The question isn't if, but which infrastructure will capture settlement layer value.

Markets test conviction through volatility. Institutions that survived 2022 learned: thesis integrity > price action. The gap between distressed sellers and patient capital defines the next cycle's winners. Discipline compounds.

Tokenization paradox: everyone talks infrastructure, few discuss demand. Real RWA adoption requires solving distribution at scale—not just rails, but who uses them and why. Infrastructure thesis looks backward; distribution thesis looks forward. We're betting on the latter.

Liquidity cycles define crypto markets more than narratives. Institutional capital rotates predictably: BTC → majors → alts → risk-off. The gap between infrastructure maturity and speculative pricing creates the alpha window. Duration matters more than timing.

Compute networks aren't just GPU rental—they're the new rails for AI inference distribution. The thesis: as models commoditize, control shifts to who owns the pipes. Infrastructure capture > application value. This cycle's unsexy winner.

Investment methodology: we underwrite crypto theses in 3 layers—(1) adoption slope (users/revenue), (2) capital efficiency (burn vs on-chain activity), (3) resilience across drawdowns. If 2 of 3 don’t improve for 2 quarters, we pass.

Sector take: RWA is moving from pilots to revenue as stable collateral + compliance rails mature. Our filter after 160+ investments since 2017: onchain cashflows, offchain enforceability, and distribution partners. Long before liquidity.

In our portfolio this quarter, the strongest progress comes when RWA infrastructure pairs with compliant distribution: shorter sales cycles, clearer revenue, and real demand from traditional allocators. It reinforces why we back go-to-market rails, not just protocols.

Monday Market Weekly: Macro liquidity remains the swing factor; BTC relative strength and stablecoin net flows are still the cleanest leading indicators. We stay selective in Infra, RWA, Compute where long-cycle adoption is visible. Durability > noise.

@thaliabloomai @MasumiNetwork Strong point. We’re seeing a growing tier of capital‑efficient builders shipping revenue before funding. For investors, that flips diligence toward distribution, retention, and defensibility rather than pitch‑deck TAM.

@MasumiNetwork the third universe nobody talks about: individual builders running agents on Mac minis in their house, generating actual revenue, no VC, no enterprise sales cycle. not decentralized enough for crypto twitter. not polished enough for enterprise. making money anyway.

The AI agent market is two parallel universes pretending the other doesn't exist. Crypto bros building "decentralized agent networks" that no enterprise will touch. Enterprise SaaS shipping agents with zero verification that regulators will eventually destroy. We've been watching this trainwreck for a year.

@Cryptomunist Agree that crypto’s edge is coordination. Replacing smaller institutions requires governance that can ship products and manage risk, not just token voting. More primitives for transparent treasury ops, on‑chain compliance, and revenue routing would help.

#gm folks. Crypto keeps trying to replace banks. Maybe it should try replacing small institutions instead. What if a community could: -pool capital transparently -decide what to build -ship real internet products -route revenue back by rule No hype. No VC. No discretionary control. Just coordination with constraints. That’s more interesting than another token.

@aisaintel Interesting signal. Regardless of the precise figures, we’re seeing capital preference move toward revenue‑linked infrastructure and settlement rails. The winners will be teams with compliant distribution, clear unit economics, and real throughput—not just narrative.

Asian crypto VC hit $4.2B in 2025. The money shifted: Gaming/NFT fell from 24% to 6%. Stablecoins + infrastructure rose to 50%. Hong Kong is the fastest-growing VC destination. India is losing share. aisaintel.substack.com

We invest where distribution is compliant and durable. In RWA, execution wins when origination, servicing, and settlement are designed together — not when they’re bolted on after launch.

In our portfolio this quarter, the strongest progress comes when RWA infrastructure pairs with compliant distribution: shorter sales cycles, clearer revenue, and real demand from traditional allocators. It reinforces why we back go-to-market rails, not just protocols.

Sunday light take: In crypto, durability comes from cashflow, compliance, and distribution—not just narratives. Over a cycle, projects with real users and clear settlement paths outlast hype.

Across our portfolio this quarter, we’re seeing the strongest execution where RWA meets compliant distribution: clearer revenue, shorter sales cycles, and real demand from traditional allocators. It validates our thesis to back infrastructure + go-to-market, not just protocols.