‘This revolution is happening in years, not decades’: Verizon CEO Dan Schulman sees AGI within 18 months—and a century of progress in a decade

Verizon CEO Dan Schulman forecast that some form of AGI could emerge within 6-18 months, followed by quantum-era capabilities roughly two years later, warning that AI-driven labor disruption could outpace institutional adaptation. Verizon has launched a $70 million AI training initiative for U.S. workers as it prepares its workforce of more than 250,000 employees and contractors for more fluid roles. Other executives emphasized AI augmentation and the continuing need for skilled infrastructure labor: Trane said AI can improve building efficiency by roughly 15% versus design baselines, while Dycom cited ongoing shortages of workers needed to build fiber networks.

Analysis

This is not a direct earnings catalyst; it is a thematic read-through whose most investable implication is that physical bottlenecks—not frontier-model capability—will determine AI monetization over the next 12-24 months. TT has the cleanest exposure through cooling, controls and retrofit demand: constrained power availability raises the value of efficiency upgrades, supporting service mix and pricing rather than merely one-time equipment volumes. DY benefits only if broadband and data-center connectivity awards convert into funded project starts; its labor-intensive model means wage inflation and execution capacity can absorb a meaningful portion of incremental revenue upside.

PCG is a second-order beneficiary of load growth, but equity capture depends on regulatory approval, allowed ROE and wildfire-liability treatment rather than electricity demand alone. The likely near-term consequence of accelerating data-center demand is higher interconnection queues and capex requirements, which can improve the rate-base outlook but also increase financing needs and political scrutiny over customer bills. VZ has limited demonstrated AI revenue sensitivity; network investment may be necessary to retain enterprise relevance without producing a commensurate return, leaving it a weaker AI proxy than infrastructure suppliers.

The article's central technological timetable is not independently investable and contains an attribution inconsistency that should be verified before assigning it informational value. Consensus may also be overextending the AI-infrastructure trade: a slowing hyperscaler capex cycle, delayed grid connections, or lower-than-expected data-center utilization would first hit cyclical installation and tool demand. Over 6-18 months, the more durable trade is efficiency and recurring service revenue over pure construction-volume exposure.

Key Decisions for Investors

• Overweight TT versus SWK on a 6-12 month horizon: TT's HVAC controls/service mix offers more direct margin capture from power-constrained retrofits, while SWK's automation narrative requires evidence of material order conversion. Reassess if TT organic orders decelerate below mid-single digits or data-center-related bookings fail to expand.
• Keep DY on a watchlist rather than initiate on this item alone. Enter only after backlog growth and project-start cadence confirm fiber/data-center connectivity demand; target a 9-12 month position with a stop/review trigger if labor costs rise faster than revenue or backlog conversion slips for two quarters.
• Use PCG only as a regulated-power optionality position, sized small, ahead of identifiable CPUC rate-case and wildfire-liability milestones rather than as an AI beta trade. Falsification is a regulatory outcome that lowers authorized returns, disallows incremental capex, or materially raises equity financing needs.
• Avoid adding VZ on broad AI rhetoric. Require evidence of incremental enterprise connectivity, edge-computing revenue, or improving return on invested capital before treating network AI demand as earnings-accretive; otherwise capex intensity remains the dominant risk.