Thoughts on the Market
2026-09-28 · Hosted by Mike Wilson · Morgan Stanley
Executive Summary
Recorded on the road from Morgan Stanley's annual Industrials Conference in Laguna Beach, this episode covers two physical-economy themes underpinning the AI buildout: autonomous trucking's shift from proving viability to commercial scale-up, and an increasingly tight power market. Transportation analyst Ravi Shanker said the industry debate has moved from "does this work" to "can this work for me," citing players like Aurora (nearly half a million driverless highway miles) and Kodiak now running revenue-generating routes for customers.
Key Stories & Changes
1. Autonomous Trucking Shifts From Proof-of-Concept to Commercial Scale-Up
Ravi Shanker: after roughly a decade of development, the industry is entering a 12-to-18-month period ahead of serial commercial production
Aurora has logged almost half a million miles of fully driverless, revenue-generating highway operations, day and night, for multiple customers
Kodiak is running several trucks in revenue service for customers including Atlas
The debate has shifted from whether the technology works to whether it can be deployed operationally at scale ("dotting i's and crossing t's")
2. Updated Cost Math: ~20% Lower Cost Per Mile, Utilization Drives Most of the Gain
Morgan Stanley's revised total-cost-of-ownership (TCO) analysis: labor accounts for 35-40% of a trucker's P&L; replacing a human driver with an autonomous system (at roughly 85 cents/mile for the autonomous technology) still nets significant savings versus the ~$3/mile labor cost today
Autonomous trucks are an estimated 13-22% more fuel efficient (Morgan Stanley used the conservative low end); additional savings come from insurance and maintenance
Net estimated saving: roughly 20% per mile versus a human driver, even after incremental costs for human drayage at route endpoints
Critically, unit economic savings account for only about one-third of total value; higher utilization drives the other two-thirds
3. Misconception Debunked: Autonomous Trucks Aren't Highway-Only
A common industry misperception is that autonomous trucks require human drivers at both ends because they can only run on highways
Morgan Stanley's AlphaWise geolocation analysis mapped over 10,000 of the largest commercial facilities belonging to the top 100 US shippers
Finding: the average commercial facility sits less than two miles from the nearest highway ramp; autonomous trucks can comfortably handle 7-10+ miles off-highway on main roads
Shanker said this points to a broader education gap in the industry that needs to close over the next 12 months before serial commercial production begins
4. Power Equipment Market: Consistently Positive, Increasingly Tight
Dave Arcaro: commentary across large-frame turbine providers and smaller on-site power equipment makers was unusually consistent and positive
Despite headlines about data center moratoriums and community pushback, power equipment companies reported no disruption to project timelines, bookings, or delivery slot reservations
GE Vernova (large-frame gas turbines) is now in conversations to contract turbine capacity for 2031 and 2032
INNIO (smaller-scale data center engines) is taking reservations into 2029 and 2030
Utilities are now planning further into the 2030s than previously typical, a new and notable shift in visibility horizon
5. Grid vs. On-Site Power: A Live Debate Among Investors
Utilities and large turbine makers argue that eventually all data center demand goes to the grid
Smaller equipment manufacturers and power-as-a-service providers counter that "nobody wants the grid," instead pursuing 15-20 year on-site power contracts directly with data center customers
Arcaro's house view: the on-site power market will become extremely large by 2030, given constraints on how much power the grid can realistically deliver to data centers
Pricing across the power equipment sector is rising and expected to keep rising into the 2030s, with no signs of softening regardless of company or equipment type
Trends Identified
1. AI's Physical Infrastructure Bottleneck Is Now Multi-Year and Structural
Both the trucking and power stories point to the same underlying dynamic: the AI economy's next phase depends less on software breakthroughs and more on physical capacity — trucks, turbines, and grid connections — that take years to build and are already being booked out into the next decade.
2. Power Equipment Sector Entering a Sustained Multi-Year Seller's Market
With turbine makers booking capacity five-plus years out and pricing still rising, the power equipment supply chain is emerging as a direct, extended beneficiary of AI CapEx (capital spending on infrastructure) — a theme distinct from, but complementary to, the chip and hyperscaler-focused AI trade dominating other coverage this week.
3. Utilization, Not Just Labor Substitution, Is the Real Automation Value Driver
Shanker's finding that two-thirds of autonomous trucking's cost advantage comes from utilization gains — not simply removing driver wages — reframes the investment thesis around operational efficiency rather than a simple labor-replacement narrative. ---
Sentiment Analysis
Overall Market Sentiment: Constructive/Bullish on AI-Adjacent Physical Infrastructure
Both analysts convey clear optimism grounded in on-the-ground conference commentary, with no significant caveats or bearish counterpoints raised in this episode.
Risk Factors Highlighted
Industry education gap on autonomous trucking: Persistent misconceptions about operational range could slow adoption timelines if not addressed.
Data center execution headwinds: Moratoriums, political pushback, and community challenges remain a headline risk even if not yet visible in equipment bookings.
Extremely long equipment lead times: Turbine orders booked into 2031-2032 mean any demand miscalculation could create significant oversupply or undersupply years later.
Grid capacity constraints: Uncertainty over how much power the grid can realistically deliver to data centers could shift more demand — and cost — onto still-developing on-site power solutions.
This episode was covered in today's [The Market Signal — 2026-09-28](https://marketsignal.beehiiv.com/p/the-market-signal-2026-09-28), a cross-source synthesis of multiple podcast reports.