Earnings Season Surprises

Tesla beat Q1 adjusted EPS at $0.41 and topped expectations for vehicle deliveries and revenue, but slightly missed energy storage deliveries while lifting 2025 capex guidance to $25 billion from $20 billion. IBM also beat estimates but kept full-year guidance unchanged, GE Vernova posted a $12 billion jump in unearned revenue with backlog near $200 billion, and Texas Instruments reported revenue up 19% with data center revenue up 90%. Progressive remains under pressure despite a 7% indicated yield, as growth slows and the insurers' cycle appears to be peaking.

Analysis

The common thread is not “AI demand” but capex pull-through into the non-glamorous layers of the stack. GEV and TXN are the cleaner second-order beneficiaries because their order books and end-markets get paid before the sexy model-training winners do; that means revenue visibility can stay strong even if hyperscaler sentiment wobbles. The market may be underestimating how much of the current infrastructure cycle is now self-funding through backlog conversion rather than fresh enthusiasm, which tends to make the down-cycle later and shallower.

Tesla’s most important signal is not the robotics storyline, it’s that management is still preserving optionality while keeping the core auto franchise profitable enough to finance it. If the company can continue generating cash without fully ramping capex, the bear case that it must choose between AI moonshots and auto economics weakens. The risk is execution timing: if the second-half spend step-up slips, the market will treat the futuristic projects as vaporware and re-rate the stock back toward a car multiple for months, not days.

IBM looks like a classic “good quarter, bad setup” situation where the market is punishing forward uncertainty more than current fundamentals. The better trade is not on whether AI helps IBM at all, but whether consulting decelerates enough to offset software reacceleration; that gap is likely to stay noisy over the next 2-3 quarters. Progressive is the opposite: the stock weakness is less about business quality and more about cycle math, with underwriting margins likely near peak and the next earnings revisions probably still drifting lower before they stabilize.

The contrarian miss is that the highest-quality names here are not the ones with the loudest AI narratives. TXN and GEV may have the best risk-adjusted setups because they benefit from broad infrastructure spend without needing a perfect monetization story. Conversely, the market may be too optimistic on near-term TSLA catalysts and too pessimistic on PGR’s long-term share gains, but both need the cycle to cooperate before multiple expansion can do the heavy lifting.

Key Decisions for Investors

• Long GEV on any 5-8% post-earnings pullback over the next 2-4 weeks; use a 6-12 month horizon. Risk/reward favors upside if backlog converts as expected, but cut if 2029-2030 booking commentary weakens.
• Accumulate TXN on weakness for a 9-15 month hold; pair it against a basket of higher-beta AI semis if you want cleaner exposure to infrastructure capex rather than compute hype. The setup improves if data-center pricing stays firm.
• Trade TSLA with a barbell: tactical call spread for the next earnings/capex update, but size modestly because the multiple is still highly sensitive to any delay in robot/terafab execution. Best entered only on confirmation of spend acceleration.
• Stay neutral-to-underweight IBM for 1-2 quarters unless you specifically want a mean-reversion trade; the core risk is not headline AI, but guidance compression if consulting weakness persists. A short-dated put spread is the cleaner expression than outright short.
• Do not chase PGR for yield; instead, wait for another 5-10% drawdown or evidence that the pricing cycle is stabilizing. Long-term, it remains a high-quality compounder, but near-term earnings revisions still look vulnerable.