AI Infrastructure Spending Set to Outpace Canal, Rail and Electrification Eras Combined

AI infrastructure spending is projected to average 3.6% of US GDP over the coming years, exceeding the combined economic weight of canal, railway and electrification investment booms. 1
A decade of asset returns shows NVIDIA +13,817% and Bitcoin +13,793% leading a broad risk-asset rally, while US bonds returned just +12% against +39% inflation. 1
Government interest costs have climbed to 3.2% of GDP, a 60-year high, while non-financial corporate interest costs sit at a 60-year low of 0.7% — evidence cited for fiscal dominance. 1
Density launched Instant, letting users query billions of rows of building sensor data at 1-second intervals and generate charts on the fly. 1
Creators demonstrated AI-driven video workflows: Opus 5.5 driving Blender, Python and After Effects for particle effects, and a Claude-assisted TikTok music video produced for $200 in 9 hours. 1 2
Clifford Asness defended tax-aware long/short strategies, arguing that without pre-tax alpha the math fails after fees and taxes, and that tax-aware hedge fund strategies date back at least 60 years to Alfred Winslow Jones. 1

AI infrastructure spending is expected to average 3.6% of US GDP over the next several years, a share larger than canal investment (1836–41), railway investment (1870–90) and electrification (1905–25) combined, according to Charlie Bilello. 1
Over the past decade, NVIDIA returned +13,817% and Bitcoin +13,793%, followed by AMD at +9,074% and Tesla at +2,625%, while the S&P 500 (SPY) gained 317%, gold 203%, US bonds (BND) 12% and US inflation rose 39%. 1
Bilello also noted that since 1980 the S&P 500 has averaged roughly 12% annual returns alongside an average intra-year drawdown of -14%, framing drawdowns as inseparable from gains. 1

Lyn Alden cited government interest expense at 3.2% of GDP — a 60-plus-year high — against non-financial corporate interest expense at 0.7%, a 60-plus-year low, as chart evidence of fiscal dominance. 1
Alden argued fiscal dominance is a spectrum rather than a binary state, and that the current environment sits clearly within it; the hard definition traders use describes a fully priced end-stage that is almost not worth defining once reached. 1

Jason praised Density's launch of Instant, which can answer utilization questions and generate charts instantly against the company's massive sensor dataset, produced and stored at 1-second intervals across billions of rows of building data. 1
Marc Andreessen amplified a demonstration using Opus 5.5 to drive Blender, Python and After Effects, converting combat footage into animated particle fields by parsing silhouettes and depth, switching particle size, color and background every half-beat, and flattening or scattering particles on impact; the creator noted that oversized particles obscure content, undersized ones lack impact, and that constant explosions make characters unreadable so bursts should be brief and weak. 1
Another creator used Claude to browse TikTok, then gave it $200 and a prompt, receiving a finished music video 9 hours later. 1

Clifford Asness pushed back on criticism of tax-aware long/short strategies, arguing that if a manager has no pre-tax alpha the after-tax, after-fee math cannot work, and that if pre-tax alpha exists, deploying it with client capital is almost certainly socially beneficial by making prices more informationally efficient. 1
Asness noted that most public research on tax-aware investing examines whether tax-aware operations reduce pre-tax alpha rather than tax law itself, and that tax-aware hedge fund strategies are at least 60 years old — Alfred Winslow Jones ran two highly levered long/short equity partnerships in 1966, one for top-bracket investors and one for lower-bracket investors, trading differently by investor tax rate. 1
Asness also questioned the claim that not every product needs to be built, asking whether a manager deciding whether and how to close a position should consider tax consequences for the client. 1

Investing