The PMI Code Says Growth. The Services Delta Says Something Else.

The composite PMI hit 56.0. Third consecutive month of expansion. The headline screams acceleration. The market will read this as a green light for risk assets. I read it as a signal divergence that most analysts will miss because they are staring at the aggregate number instead of the underlying code. Here is the raw data from the S&P Global release. Services PMI jumped 2.2 points to 56.8. That is the highest print since March 2022. Manufacturing, meanwhile, fell 0.7 points to 53.9. That is a five-month low. The spread between the two is now nearly three full points. That is not a minor wobble. That is a structural fracture in the economic engine. Let me be precise about what this means. The composite index is a weighted average. When the services component surges while manufacturing stumbles, the composite masks the internal tension. The chart is a symptom, not the cause. The cause is a two-speed economy where AI capital expenditure is inflating the services sector while interest-rate-sensitive manufacturing bleeds momentum. I have been running market surveillance long enough to know that the first derivative matters more than the level. The services PMI is accelerating. Manufacturing is decelerating. That divergence is the signal. The composite is just the noise filter that obscures it. Now, the context. This is not a normal cyclical expansion. The report explicitly attributes the growth surge to AI. That is a bold claim, and my code-first verification habit demands I stress-test it. Based on my audit experience, when a narrative attributes growth to a single technological factor, I look for the transmission mechanism. Here, it is clear. AI is not a manufacturing story. It is a services story. Software, cloud infrastructure, data analytics, financial services, legal research. That is where the productivity gains are landing first. The hiring data confirms this. The report notes that employment growth accelerated to its fastest pace since January 2025. That is a hard data point, not a narrative. When services firms are hiring at that clip, it means the demand signal is real. It means the AI-driven expansion is translating into payrolls, which translates into income, which translates into consumption. The feedback loop is intact. But here is where I diverge from the consensus read. The mainstream interpretation will be that this is a clean acceleration story. Strong growth, strong jobs, AI tailwinds. Buy the dip. Extend risk. I see a different code running underneath. The manufacturing weakness is the canary. A 53.9 print is still above the 50 boom-bust line, but the trend is unmistakable. This is the third consecutive month of manufacturing deceleration. Historically, when manufacturing leads the downturn while services lag, it signals a policy transmission problem. The Fed's tightening cycle, even if paused, is still working its way through the interest-rate-sensitive corners of the economy. Manufacturing is feeling it. Services, buoyed by the AI capex wave, is not. This is not a sustainable equilibrium. Either the AI-driven services boom pulls manufacturing along with it, or the manufacturing weakness eventually drags services down. The composite PMI of 56.0 implies a Q3 GDP print of around 3.0 percent, double the 1.5 percent from Q2. That is a massive acceleration. But I am skeptical of the sustainability. Let me run the numbers through my historical mapping model. A composite PMI of 56.0 typically corresponds to annualized GDP growth between 2.5 and 3.5 percent. The 3.0 percent forecast sits at the upper end of that range. That is achievable, but it requires the services sector to maintain its current trajectory. And that requires AI capex to keep flowing at current levels. Here is the contrarian angle that nobody is talking about. The market is pricing this as a pure growth story. But the composition of that growth matters more than the headline. Services-led growth with manufacturing lagging is a lower-quality expansion. It is more dependent on financial conditions remaining loose. It is more vulnerable to a sentiment shift in the AI trade. And it carries a hidden inflation risk. Services PMI at 56.8 with accelerating hiring is a classic recipe for core services inflation. Wages are the stickiest component of the inflation basket. When services firms are hiring aggressively, they are bidding up labor costs. That eventually feeds into CPI. The report does not mention inflation, but the data is screaming the implication. If Q3 GDP does come in at 3.0 percent, the output gap closes, and the Fed's room to cut rates evaporates. This is the policy trap. The market is still pricing in a rate cut this year. But the data is moving in the opposite direction. Strong growth, accelerating hiring, and a services sector running hot. That is not a recipe for easing. That is a recipe for the Fed to hold rates higher for longer. And if the AI-driven growth narrative holds, the Fed might even have to discuss the unthinkable. Not cuts. Hikes. Let me be clear about the transmission mechanism. The Fed's reaction function is data-dependent. The data is now showing an economy that is accelerating, not decelerating. The case for preventive cuts is dead. The case for patient观望 is alive. And if the September PMI comes in above 55 again, the market will have to reprice the entire rate path. Now, the institutional due diligence angle. I have been tracking the AI capex cycle since the beginning. The current wave is different from the 2021 crypto bubble. Back then, the capital was flowing into speculative assets with no cash flows. Now, the capital is flowing into productive infrastructure. Data centers, chips, power grids, cooling systems. That is real investment. But it is also a massive bet on future returns. If the AI buildout does not generate the expected productivity gains, the capex cycle will reverse. And that reversal will hit the services sector hardest. The manufacturing-services divergence is the early warning. Manufacturing is the canary in the coal mine because it is the most interest-rate-sensitive sector. It is telling us that the economy is not as strong as the composite suggests. The services strength is real, but it is concentrated in the AI complex. That is a narrow base for a broad-based expansion. Let me give you the trade implications. The market will rally on this data. The AI trade will extend. But the smart money should be watching the September PMI release. If the composite holds above 55, the acceleration narrative is confirmed. If it drops below 54, the divergence becomes a convergence, and the downside risk increases. I am also watching the Q3 GDP print. If it comes in at 3.0 percent, the market will have to accept that the Fed is on hold indefinitely. That is a headwind for long-duration assets. If it comes in below 2.0 percent, the entire growth narrative collapses, and the market will pivot to recession pricing. The range of outcomes is wide, and the data is not yet decisive. Here is my takeaway. The composite PMI is a headline. The services-manufacturing spread is the code. And the code is telling me that this expansion is narrower than it appears. The AI-driven services boom is real, but it is not broad-based. The manufacturing weakness is a warning that the transmission mechanism is broken. The Fed is trapped between an accelerating services sector and a decelerating manufacturing sector. The policy response is unclear. The market is pricing a smooth path. I am pricing volatility. Sleep is for those who can. I will be watching the September data releases with a forensic eye. The next PMI print will tell us whether this is a new growth cycle or a head-fake. The code does not lie. The narrative does. Signal over noise. Always.