Gutenberg Lab — Morning Paper

Crypto Splits Three Ways on Quantum Defense as Bitcoin's 5M Unmigrated Coins Threaten Market-Wide Trust Collapse

Speakers

Laura Shin (Unchained Host) × Charles Guillamey (Ledger CTO) × Ilya Polosukhin (Near Protocol Co-founder)

Bottom line

As quantum computing threats compress from theoretical to imminent, the crypto industry has fragmented across three post-quantum algorithms, with Bitcoin's governance paralysis over Satoshi's ~5 million unmigrated coins emerging as the single greatest risk to industry-wide trust—worse than any technical vulnerability.

3-Line Summary

• NIST standardized three post-quantum algorithms in 2024, but crypto split into lattice-based (Near/Algorand), hash-based (Ethereum/Bitcoin), and speed-first (Solana) camps, creating coordination nightmares for hardware wallets and cross-chain infrastructure
• Google's March 2024 zero-knowledge proof of an improved Shor's algorithm triggered EigenLayer's AI competition to find and beat it within 48 hours, compressing the quantum timeline from "five years away" to "accelerating with massive investment and AI"
• Bitcoin's ~5 million Satoshi-held coins face three options—let them be stolen, freeze them permanently, or freeze and tail-emit until 2176—but failure to reach consensus before quantum computers arrive could trigger hard forks and market collapse

Three Key Points

1. AI Compressed Quantum Timeline in 48 Hours: EigenLayer's AI competition found and beat Google's secret Shor's algorithm improvement within 48 hours of launch, with continuous improvements since. This demonstrates AI's power for systematic algorithm optimization and why "five years away" now feels qualitatively different from past decades of similar predictions (Ilya, 00:04:00).

2. Hardware Wallets Weren't Designed for Post-Quantum: Current secure elements were built for banking/identity applications and lack hardware-level support for post-quantum algorithms. Ledger must implement at firmware level with lower performance and reduced security properties until next-generation chips arrive—a multi-year hardware replacement cycle overlapping with the quantum threat timeline (Charles, 00:29:16).

3. MPC Capability Divides Algorithm Choices: Multiparty computation (MPC) is "most likely the only algorithm" possible with lattice-based MLGSA, and "very unlikely to be possible" with hash-based signatures. This matters enormously for institutional adoption and cross-chain infrastructure, creating pressure for ecosystem convergence despite Ethereum/Bitcoin's hash-based preference for conservative security (Charles, 00:24:29, 00:27:35).

Editorial Perspective

Bitcoin's governance paralysis is transforming a technically solvable problem into an existential threat. As Charles notes, the entire non-blockchain world settled on MLGSA with no debate, while crypto fragments across three algorithms—technical implementation is "a little bit easier for blockchain," yet social consensus is harder. This mirrors the structural failure of early internet platforms when faced with disruptive technology shifts: the technical migration path existed, but coordination failure across stakeholders killed the platform before external shock arrived. Ilya's warning that "Bitcoin attack would bring all of crypto down—crypto's built on trust in technology" captures the platform transition dynamic: trust erosion propagates faster than individual code security. The quantum computer's arrival date remains uncertain, but the governance deadlock is already measurable—and more dangerous than the quantum threat itself.

Source: Unchained "One Type of Post-Quantum Cryptography Is Most Popular. Why Is Crypto Trying Out Three?" (August 25, 2026)

https://unchainedcrypto.com

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

Open-Weight AI Expands the Market, Not Splits It

Featuring: Theo Jaffe × Aaron Levie (Co-founder & CEO, Box) × Sofia Puccini

Bottom line:

Open-weight AI models are not a zero-sum threat to closed frontier labs but rather expand the total AI ecosystem by enabling new use cases and ultimately funneling revenue to infrastructure providers through inference costs, regardless of model openness.

Three-line summary

Box CEO Aaron Levie rejects the framing of open-weight models as zero-sum with closed models, arguing they expand the market by creating new use cases. In response to China's progress (e.g., Kimi K3), he calls for the U.S. to invest $10 billion in open labs. He predicts the future of enterprise AI will center on model routing layers that abstract away provider lock-in.

Three key points

1. Distillation is economically defensible: If AI models can train on the public internet, other AI models can train on AI outputs—Levie sees no ethical line crossed. In fact, Anthropic earns more from distillation API usage than most labs earn from original training runs.

2. Blocking China accelerates their independence: Levie agrees with Nvidia CEO Jensen Huang that China has the talent, industrial capacity, and strategic commitment to build AI regardless of U.S. restrictions. Cutting them off will force them to build a parallel stack that competes globally. "I just don't think there's a scenario here where you can kind of close off completely to China and somehow... we just win."

3. Inference revenue trumps model IP: In a market with 3-5 competing U.S. labs, token costs will converge toward infrastructure costs (20-40% margin). Therefore, whoever powers inference, post-training, or enterprise trust captures revenue regardless of whether the model is open or closed.

Editorial perspective

Levie's argument exemplifies the "separation of standardization and differentiation" pattern in platform transitions. During the early internet era, the browser wars ultimately settled through the separation of HTTP as a standard layer from the application layer. The same structure is emerging in AI—models themselves become standardized, while value migrates to inference infrastructure and task-understanding layers. Box's positioning as an "applied AI layer" anticipates this separation. However, as Anthropic's safety-first stance demonstrates, standardization is not a technological inevitability but a political choice. Who decides the risk framework? The answer to this question will redraw the boundary between open and closed.

Source: a16z "Aaron Levie on Why Open AI Wins" (September 5, 2026)

https://a16z.simplecast.com/episodes/aaron-levie-on-why-open-ai-wins-YUnuv_hR

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --

Perps Classification Fight: CME vs. CFTC Battle Determines Retail Access to Innovative Derivatives

Featuring

Laura Shin (Unchained host & moderator) × Tiffany Smith (Partner, Wilmer Hale) × Jake Shervinsky (Hyperliquid Policy Center) × Kathy Yoon (Temporal)

Bottom line

Classifying perpetual futures as "swaps" restricts them to institutions; classifying them as "futures" opens retail access—this legal distinction determines whether ordinary Americans can trade an innovative derivative.

3-Line Summary

• Chicago Mercantile Exchange (CME) is suing the CFTC, arguing perpetual futures ("perps") should be classified as swaps, not futures—a distinction that gates retail access
• In July 2026, the CFTC authorized the first perpetual futures for Coinbase and Kalshi; simultaneously, Hyperliquid is reportedly negotiating US onshoring through Kraken's Bitnomial subsidiary
• The panel warns that forcing regulations designed for 20th-century bilateral contracts onto autonomous smart contracts misses both risks and opportunities

3 Key Points

1. Regulation by enforcement created the confusion

Tiffany Smith reveals that five CFTC enforcement actions classified perps as swaps, causing the industry to believe that view. Now, for the first time, a policy division (not enforcement) has classified them as futures—"another example of what happens where you have regulation by enforcement."

2. Perpetuals are structurally superior

Jake Shervinsky cites three advantages: (1) no fragmented liquidity across expiration dates, creating deeper liquidity; (2) simpler to use because they closely track the underlying asset price, unlike dated futures or options; (3) proven benefits in metals (silver, gold), oil, and other markets where 24/7 trading is valuable.

3. Hyperliquid is infrastructure, not an exchange

Jake emphasizes: "It is a mistake to think of Hyperliquid as an exchange. Hyperliquid is infrastructure that any exchange can use." The regulatory challenge is "how can core principles be satisfied where the exchange offers a product using neutral infrastructure that they themselves do not control and also no single third party controls."

Editorial Perspective

Tiffany's observation—that the 2000 Commodities Modernization Act and 2010 Dodd-Frank were "created for totally different products"—echoes the structural mismatch when regulators tried to apply feature-phone content rules to the smartphone app economy. The purpose of regulation may remain constant, but when the technological substrate changes, the means must be rebuilt. The perps classification fight is a case study in incumbents resisting that rebuild.

Source: Unchained, "Why the Question Over How to Regulate Perps Has Turned Into a Fight" (September 4, 2026)

https://unchainedcrypto.com

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --