The macro picture deteriorated modestly on Thursday. Initial jobless claims came in at 228K — above the 220K forecast and the third consecutive week of rising claims — adding a subtle softening of the labour market signal to the day’s geopolitical and earnings pressures. More consequentially, April’s S&P Global flash manufacturing PMI printed 48.7, below the 50-expansion threshold for the second month in a row, with the services PMI at 51.4 narrowly holding above contraction. The combination of a contracting manufacturing sector, $103 Brent crude, and a Fed on hold suggests stagflation risks are re-entering analyst vocabulary — precisely the macro environment Kevin Warsh warned about in his Senate confirmation testimony.

The 10-year Treasury yield edging up to 4.33% on a risk-off day is the most telling detail from the fixed income market. Normally, equity selloffs are accompanied by a Treasury rally (yields falling). The fact that yields rose alongside the equity decline signals that the selloff is stagflationary in character — the market is pricing higher inflation (via oil) and reduced growth simultaneously, with no Fed cut as relief. This is the most dangerous macro configuration for equities: neither bonds nor rate cut expectations are providing a cushion. The 4.40% 10-year level — which sparked a significant equity correction earlier this month — is now only 7 basis points away.

Thursday’s earnings tableau tells two stories simultaneously. The first story is hardware and real assets: Texas Instruments, United Rentals, and Comcast all delivered genuine beat-and-raise results, confirming that physical infrastructure, semiconductor hardware, and live-event media are thriving in the current environment. TXN’s +15.4% is the single best earnings move of the week and extends the semiconductor sector’s extraordinary 17-session streak — a signal that AI hardware investment is accelerating, not decelerating. The second story is enterprise software and AI narrative names: IBM, ServiceNow, and Tesla’s capex reset are a unified signal that the AI revolution is actively disrupting traditional software revenue models faster than new AI-native revenue arrives. The question for next week’s Alphabet, Microsoft, Meta, and Amazon prints is whether the cloud and AI infrastructure layer (AWS, Azure, GCP) confirms the TXN hardware thesis — or whether the IBM/NOW enterprise disruption signal is the correct read on the broader AI transition.

The Tesla post-earnings story is a masterclass in market psychology. An EPS beat of $0.41 against a $0.37 consensus — an 11% beat — should, in isolation, be a positive catalyst. Instead, the stock fell −3.33% because the market is not valuing Tesla on trailing EPS but on its ability to fund a multi-year AI and autonomy transition without destroying free cash flow. Raising full-year capex to $25B — a level exceeding Tesla’s total free cash flow generation from the last three years combined — fundamentally challenges the self-funding narrative that underpins Tesla’s premium valuation.

The energy segment’s −12% revenue decline is equally alarming. Tesla’s energy storage and solar business was supposed to be the second growth engine as EV growth moderated — its contraction in a quarter where AI datacenter power demand is at all-time highs suggests Tesla is losing market share in a market it should be winning. Dan Ives of Wedbush has long argued that Tesla is “more an AI company than a car company” — but Thursday’s results challenge that thesis on both sides: the car business is losing ground to BYD and Xiaomi globally, and the AI/energy business is not yet filling the gap. The path forward requires Robotaxi to work at commercial scale in new cities by Q3, and Optimus robots to generate meaningful revenue by year-end. Both timelines remain entirely speculative.