.png)
The most striking signal is the one that is largely absent. References to economic slowdown peaked at around 810 per month during the 2022 growth scare; today the count is 74 — less than 10% of that peak — and has remained broadly within a 45–75 range for the past year. The latest increase is concentrated in Health Care, and excluding that sector slowdown references fell to a six-month low. There is, in short, no broad demand-warning signal in management language.
That reading comes with a caveat. Management commentary is primarily a coincident rather than a leading indicator of demand — companies rarely flag a slowdown before it becomes visible in orders — so it should be read as an absence of current deterioration, not a forecast that weakness cannot emerge.
Chart 1 · Economic slowdown mentions by GICS sector
Source: Syz Research / Bloomberg
The individual transcripts illustrate what the aggregate signal is capturing. During the Q2 reporting season, PNC CEO Bill Demchak described growth as too broad-based to attribute solely to AI; US Bancorp reported growth across almost every commercial category, while Honeywell recorded broad-based order growth across its short-cycle businesses.
There are pockets of consumer weakness — Kroger, Nike and Hershey highlighted pressure on lower-income households — but Visa reported no weakening across lower spending bands and JPMorgan continued to describe consumers and small businesses as resilient. The message is therefore one of dispersion rather than broad demand contraction.
Inflation tells almost the opposite story. Management references fell steadily from more than 2,500 per month in 2022 to just 410 in January 2026, then almost tripled to 1,171. More importantly, the increase has been broad: eight of eleven GICS sectors are higher over the past three months, led by Industrials, Consumer Discretionary, Materials and Consumer Staples.
Chart 2 · Inflation mentions by GICS sector
Source: Syz Research / Bloomberg
The supply-chain series helps explain the source. Supply-chain disruption references rose from 221 in January to 535 in May, but the composition of the 2026 episode is very different from the pandemic. In 2021–22, the problem was largely freight, logistics and consumer goods; this time, Information Technology and Industrials dominate. The bottlenecks are increasingly chips, memory, electricity, grids, equipment and productive capacity.
Chart 3 · Supply-chain disruption mentions by GICS sector
Source: Syz Research / Bloomberg
Management commentary reinforces that interpretation. Microsoft said customer demand continued to exceed available capacity; Qualcomm described parts of the semiconductor industry as operating near full utilisation, while Micron expects DRAM and NAND demand to remain above supply beyond 2027. The pressure is spreading into the physical economy: Duke Energy has secured 7.8 gigawatts of electric-service agreements with data-centre customers, Chevron says grid supply cannot keep pace with hyperscaler demand, and Commercial Metals sees a multi-year construction pipeline across data centres, semiconductor capacity and energy networks.
The distinction is central. The 2026 inflation impulse has not primarily come from collapsing supply hitting a weak economy; it has come from demand growing faster than available capacity. That is initially constructive for earnings — high utilisation supports investment, backlogs and pricing power — and the risk emerges only later, when higher costs begin to pressure margins. As our recent Focus argued, scarcity created by strong demand remains supportive until input-cost inflation begins to overwhelm pricing, productivity and operating leverage.
The August data suggest the sequence may now be turning. Supply-chain disruption references have fallen for three consecutive months to 387 — 28% below their May peak — with ten of eleven sectors declining, while inflation references have only just recorded their first monthly decline. That creates a potentially useful sequencing signal: management discussion of the upstream problem may be turning before discussion of its downstream inflation consequences.
The three-month gap is suggestive, but one episode is not enough to establish a stable lead-lag relationship. The value of the dataset is that this can now be tested systematically against inflation, margins and earnings revisions.
Job-cut references reinforce the benign interpretation, falling from a spring plateau of around 145–148 to 100 in August, also declining for three consecutive months.
Chart 4 · Job-cut mentions by GICS sector
Source: Syz Research / Bloomberg
The remaining references look more structural than cyclical, particularly in Software & Services, consistent with restructuring and AI-related efficiency rather than a demand-driven firing cycle. There is no broad labour-shedding signal in current management commentary.
AI is by far the dominant corporate narrative. AI and machine-learning references reached 6,115 in August, up 144% year on year and roughly five times the inflation series. But the most interesting development is who is doing the talking.
Chart 5 · AI / machine-learning mentions by GICS sector
Source: Syz Research / Bloomberg
Information Technology’s share of AI references has fallen from around 62% two years ago to roughly 52% today, and outside IT mentions have approximately doubled since January. AI is moving from a vendor narrative to an adopter narrative. The company evidence supports that diffusion: Cisco reported USD 9.3bn of AI-infrastructure orders for the year and guided to USD 7.5bn of AI-infrastructure revenue in FY27, CoreWeave ended Q2 with a USD 104bn backlog, rising to around USD 129bn by mid-August, and Applied Materials reported record revenue as DRAM, leading-edge logic and advanced-packaging orders accelerated. At the same time, Duke Energy, Chevron and Commercial Metals show how the investment cycle is spreading into power and physical infrastructure.
August adds an important nuance. Excluding Information Technology, AI references fell 15% in August and are now 17% below their May peak.
Chart 6 · AI / machine-learning mentions excluding Information Technology
Source: Syz Research / Bloomberg
The sectors that led the adoption surge — Financials, Industrials, Health Care and Communication Services — are now leading the decline. This does not imply that adoption is reversing: ex-IT references remain around twice January levels. A more interesting interpretation is that AI “talk intensity” among adopters may have peaked.
If so, the next phase becomes measurable. Companies have spent several quarters discussing pilots and deployments; investors should now increasingly expect evidence in productivity, revenues, margins and capital efficiency. That gives the tracker another potential use: distinguishing a genuine transition from narrative to execution from simple narrative fatigue.