Conference Presentation, Interview, Webinar
AI, Infrastructure, and the Next Investment Cycle
- Eight of the top 10 global companies by market value are U.S. tech firms, driving a market rise of 90% (17% annualized) over the four years since Sanjay Chattopadhyay's 2020 announcement.
- Global infrastructure investment needs are estimated at $90 trillion through 2040, extending beyond AI data centers to include power, water, roads, and transit.
- High-tech equipment, software, and R&D now account for approximately 55% of U.S. capital spending, surpassing the railroad build-out as a percentage of GDP.
- Hyperscaler capital expenditure (CapEx) is projected to rise from $416 billion in 2025 to $780 billion in 2026, with annual spending expected to exceed $1 trillion starting in 2027.
- The S&P 500 earnings multiple remains below 20x, indicating market gains are driven by fundamental earnings growth rather than valuation expansion, contrasting sharply with the dot-com era.
- Live AI deployments at S&P 500 companies stand at 69%, but only 30% of those deployments show quantifiable impact, while the ultimate adoption metric tracks at just 2%.
- Power users in the AI sector spend roughly eight times more than the top 10% of users, with the top 1% spending nearly as much as the 2%–10% combined.
- Chime reported a compounded 50% reduction in cost-to-serve over four years due to AI, while Shopify saw an 8% increase in customers reaching five orders within 15 days of onboarding.
- ServiceNow reported over $1 billion in AI Annual Contract Value (ACV) and a 9x increase in agentic deployments.
- Only ~2% of U.S. households hold a paying AI subscription, compared to 70 million for Netflix and over 200 million for Amazon Prime, indicating significant consumer growth potential.
- Public software mix has shifted, with 75% of sampled companies profitable but only 30% growing at 20% or more annually.
- In the private market, the top six companies (Anthropic, OpenAI, Databricks, Stripe, Waymo, Revolut) hold a combined valuation of $2.4 trillion, exceeding the total market cap of all IPOs in the last decade (excluding SpaceX).
- Secondary market tenders now show near-zero discounts to the last round price, signaling renewed investor confidence compared to the 2021–2023 correction period.
- 58% of Carta employees declined liquidity in recent tender offers, reflecting high conviction in company performance rather than an exodus.
- AI-related companies accounted for 86% of U.S. VC deal activity in the 2026 snapshot, up from 65% in 2025.
- Hyperscalers like Microsoft, Google, and Amazon hold a combined cloud backlog of $1.7 trillion, with demand currently outstripping supply across nearly all segments.
- Data center capacity expansion is predicted to lower residential electricity rates by 40 basis points for every 10% increase in capacity due to shared grid cost spreading.
- Agents are driving a 14x growth in token usage on OpenRouter, with optimization techniques like caching reducing financial workload costs by 10x for firms like Hebia.
- Public market software performance is bifurcated, with cybersecurity and vertical software outperforming horizontal applications as AI creates new security monitoring needs and automates general workflows.
- Forward-looking statements indicate autonomous driving networks could expand by an order of magnitude, potentially replacing 17 million new annual U.S. car sales with autonomous vehicles over the next decade.
- Founders of private giants are advised to delay IPOs to maintain flexibility for long-duration, high-swing bets that face greater scrutiny in public markets.