1Q26_ Quarterly Outlook Report_Final_EN - Flipbook - Page 32
T H E P LUMB LI N E | A RETU RN TO F I RS T PRI N CI PL ES
Semiconductor companies have seen greater margin improvement, while the largest U.S.
firms, many of which are AI-adjacent, have accounted for the bulk of returns in recent years
100%
Annual performance contribution
Operating margin chg (4Q19 to 3Q25)
3.1%
1.4%
35%
34%
65%
66%
46%
75%
50%
54%
25%
0%
S&P 500
Semis industry
2023
2024
Top 10
Rest of index
2025 YTD
Source: Bloomberg Finance LP, Scotia Wealth Management
Phase 2: Diffusion
The second phase is the diffusion phase, where falling implementation costs broaden adoption
across the economy. As access to generative AI models becomes cheaper, following a trajectory
similar to historical declines in the cost of computation, barriers to adoption fall. Productivity
gains begin to spread beyond early beneficiaries such as semiconductor and infrastructure firms.
Lower implementation costs broaden technological diffusion
100
10
1
0
1
2
3
4
5
Years
Cost of accessing GenAI models (quality adjusted price)
6
7
8
9
10
Cost of computation (USD per instr./second)
Source: John C. McCallum (2023); U.S. Bureau of Labor Statistics (2024); OECD; Andre et al, 2025; Scotia Wealth Management.
As AI tools become more accessible, productivity improvements are likely to lift corporate
profitability across a wider range of sectors, including finance, energy, healthcare, manufacturing,
and retail. The relationship between margin expansion and earnings growth is well established. A
simple rule of thumb based on historical data suggests that a one percentage point increase in
operating margins translates into roughly an 8ppt increase in earnings per share. This relationship
was clearly visible during the Information Age, when a roughly 1.5ppt rise in operating margins
31
