US Fiscal Gap Widens, Long-Duration Bonds Sink, and Bittensor's Decentralized AI Model Hits a Milestone

Key Takeaways

1. The US fiscal picture keeps deteriorating: FY2026 revenue is $4.5T vs $6.3T spending in the first 10 months, and federal debt has doubled to $39T in a decade. — via 1
2. Long-duration bond ETFs are still 63% below their March 2020 peak as the 30-year yield rose from 0.8% to 5.3%; even the Fed's rate cuts did not stop the long end from backing up. — via 1
3. Bittensor's Subnet 3 delivered the first 10B pretrained model, Teutonic-I, which beat 40-70B decentralized models in benchmarks — a milestone for decentralized AI. — via 1
4. Raoul Pal argues banks and asset managers are moving onto blockchain rails, AI agents are proliferating, and the right move is to hold a basket of high-quality L1s instead of overthinking. — via 1
5. Global deep value stocks have outperformed the S&P 500 over the last 1, 3, and 5 years, but investors won't care until the index stops delivering ~15% annual gains. — via 1
6. The debate is shifting from "is AI a bubble?" to AI's impact on the federal budget, while David Sacks' "AI is not nuclear" frame gains traction against regulation. — via 1 2

1. US Fiscal and Bond Market Stress

1. Federal finances are far from balanced. In the first 10 months of FY2026, revenue was $4.5 trillion and spending $6.3 trillion, a gap that makes "balanced budget" claims untenable. — via 1
2. The debt trajectory has accelerated. Taxes rose 65% to $5.3 trillion and spending 96% to $7.3 trillion over the past decade, while total federal debt roughly doubled from $19 trillion to $39 trillion. — via 1
3. Long-duration bonds remain in a bear market. Long-duration bond ETFs are still down 63% from their March 2020 peak as the 30-year Treasury yield climbed from a 0.8% low to about 5.3%. — via 1
4. The post-cut bond selloff is a red flag. After the Fed began cutting in September 2024, the 30-year yield rose from below 4% to 5.3%, highlighting that inflation pressure can persist even as the Fed pivots. — via 1
5. Geopolitical positioning has flipped. Bilello highlights the contrast between Trump's 2024-25 "no wars, end wars, America First" posture and his 2026 claim that the Strait of Hormuz should become US territory. — via 1

2. Markets and Portfolio Implications

1. The "just hold the basket" case for L1s. Raoul Pal sees banks and asset managers moving products onto blockchain rails, AI agents being built everywhere, and liquidity starting to flow; with a "Clarity Act" expected and geopolitical noise fading, he says buying high-quality L1s and holding them is the right approach — the only way to lose is overthinking. — via 1
2. Deep value is winning quietly. Global deep value stocks have beaten the S&P 500 over the last 1, 3, and 5 years, but this outperformance is ignored as long as the index keeps rising ~15% a year. — via 1
3. Ackman's tech reshuffle is being read as valuation-driven. Per Jason, Bill Ackman sold Alphabet, trimmed Amazon, and added Meta, Uber, and Netflix — a set of moves that Jason likens to "finding price dislocation" rather than a broad tech call. — via 1
4. A short-seller warning remains unspecific. MuddyWaters Research calls an unnamed asset "the most obvious manipulated asset trading price" in history and shares a supporting link; the lack of asset details leaves this as an unverified red flag. — via 1

3. AI, Fiscal Policy, and Regulation

1. Bittensor's 10B model breakthrough. Teutonic-I, the first 10B pretrained model on Subnet 3, beat 40-70B decentralized models in benchmarks; Jason predicts Bittensor's native models will remain in the top 20 on cost-adjusted performance for 12 months, with a non-zero chance of top 10 in 24 months, and encourages spare-compute participants to join for AI sovereignty and token rewards. — via 1
2. AI's fiscal impact becomes the key question. Meb Faber argues that instead of arguing whether AI is a bubble, investors should focus on what AI will do to the federal budget, highlighting Luke Gromen's discussion on the topic. — via 1
3. "AI is not nuclear" pushes back on regulation. David Sacks' argument — retweeted by Jason — frames AI as consumer technology, not military technology, and treats the AI-as-nuclear analogy as a pretext for overregulation; Jason adds that sci-fi like Cyberdyne has polluted public perception. — via 1
4. Waste-cutting before wealth taxes. Jason argues that before new billionaire wealth taxes, the government should eliminate 10-30% of fraud and waste or open services to private competition; managers who tolerate large-scale fraud shouldn't receive more funding. — via 1