The week's unifying thread was the widening gap between AI ambition and AI accountability. Big Tech's earnings made clear that investors will no longer reward revenue beats alone — they want evidence that the hundreds of billions flowing into data centers will pay off, a pressure visible in Meta's layoffs, Tesla's capex jump, and Google's decision to monetize its TPUs by selling them outright. Against that backdrop, healthcare offered the week's counter-narrative: under political and provider pressure, payers loosened prior authorization and won a Medicare rate bump, trading some control for goodwill. The common question across both stories is the same one Wall Street is now asking everyone — when does all this spending actually produce a return?
Alphabet, Amazon, Meta, and Microsoft all reported within an 80-second window on April 29, with Apple following April 30 — and every one beat revenue estimates. The reactions diverged sharply: Alphabet jumped as much as 6% after raising AI infrastructure spend, while Meta fell more than 5% on flat Q2 growth guidance and Amazon and Microsoft each slid roughly 3%. The quarter confirmed that revenue beats no longer carry stocks; the market is now grading the return on AI capex.
TechnologyGrowth
Alphabet confirmed it will begin selling its Tensor Processing Units to select third-party customers for installation in their own data centers — the first time Google's custom AI silicon leaves Google's own cloud. The move gives enterprises a credible non-NVIDIA option for training and inference, and reframes Google as a direct hardware competitor rather than just a cloud rental shop. It is the most strategically consequential vendor move of the week.
AI ToolsStrategy
Tesla posted adjusted EPS of 41 cents against 37 cents expected on Q1 revenue of $22.39 billion, narrowly missing the top line as the core auto business kept struggling. Deliveries of 358,023 vehicles trailed estimates and the company built over 50,000 more cars than it sold. The headline was the spending plan: capex will now top $25 billion this year, $5 billion above prior guidance, as Tesla pours money into AI, robotics, and Robotaxi expansion now live in Austin, Dallas, and Houston.
CarsStrategy
UnitedHealthcare said it will eliminate prior authorization requirements for roughly 30% of medical services that previously needed insurer sign-off, with the changes phased in by the end of 2026. The cuts cover certain outpatient surgeries, diagnostic tests such as echocardiograms, and some outpatient therapies and chiropractic care. It is the clearest concession yet from a major payer to mounting provider and political pressure over administrative friction.
Healthcare Payers
CMS agreed to raise Medicare Advantage payments for next year, sending UnitedHealth and other insurer stocks higher on the news. The rate bump lands as payers like UnitedHealth, Humana, Elevance, Aetna, and Cigna are already aggressively repricing premiums and narrowing networks to recover margin. After a punishing stretch of elevated medical costs, the government's payment signal hands managed care a measure of relief heading into 2027 bid season.
Healthcare Payers
The FDA approved Axsome Therapeutics' AXS-05, marketed as Auvelity, for agitation associated with dementia due to Alzheimer's disease on April 30. The decision expands the drug well beyond its original major depressive disorder indication into one of the most underserved symptoms in dementia care. The same week, an FDA panel voted against AstraZeneca's camizestrant combo in HR-positive breast cancer, a reminder the agency's bar remains uneven.
Biopharma
Mark Zuckerberg told roughly 8,000 Meta employees that their jobs were a casualty of the company's AI infrastructure budget, framing the layoffs as a direct trade of headcount for compute. Microsoft is moving in parallel, expected to offer buyouts to about 7% of its U.S. workforce by the end of June. The pattern is now explicit across Big Tech: labor is being liquidated to fund data centers.
OperationsAI Tools
DeepSeek previewed V4-Pro and V4-Flash, its first major architecture refresh since V3 and the most consequential open-weight release of the year. V4-Pro runs 1.6 trillion parameters (49B active) and V4-Flash 284B parameters (13B active), both with a 1M-token context window and an MIT license that permits commercial fine-tuning. The release keeps pricing and openness pressure on the frontier labs just as their capex bills explode.
AI ToolsTechnology