Tim Cook Steps Down, Vertiv Confirms the Backlog Was a Floor, and Intel Says It Might Stop Making Chips

The Cook handover and what ‘hardware-first’ means

Apple disclosed on Monday 20 April that Tim Cook will transition from CEO to Executive Chair effective 1 September 2026. John Ternus, Senior Vice President of Hardware Engineering since 2021 and at Apple since 2001, becomes CEO. Art Levinson moves from Chair to Lead Independent Director. Cook stays on the board.

This is the first CEO transition at Apple since 2011 — fifteen years of Cook. The interesting question is not the timing (Cook is 65 and has telegraphed this transition for three years) but the choice. Apple’s senior bench had three credible internal candidates: Jeff Williams (operations), Eddy Cue (services), or Ternus (hardware). Williams would have signalled continuity-of-management. Cue would have signalled services-led monetisation, the model that has driven Apple’s revenue diversification since Cook took over. Ternus signals hardware re-prioritisation.

Ternus is 50 years old. He runs the team that designed Apple Silicon, Vision Pro, the foldable iPhone roadmap (still unannounced but heavily reported), and the on-device AI architecture for Apple Intelligence. His CEO appointment is, in a single decision, an institutional declaration that the next decade of Apple is hardware-led. Cars, services, AR/VR — these will be evaluated through a hardware-economics lens.

The transition has six months of overlap with Cook still in the building. The first major decision Ternus makes — the first guidance call, the first product cycle pre-announcement — will be obsessed over. Until then the market should treat the disclosure as institutional continuity in tone, hardware re-orientation in substance.

Vertiv prints the floor, Intel discloses the trapdoor

Issue 01 closed with three test cases for the earnings wave: VRT, DLR, and INTC. Two of the three reported with substance this week, and the contrast is the story of the week.

Vertiv confirmed the thesis explicitly. Q1 net sales $2.65B, up 30.1% year-over-year. Gross margin expanded 400bps to 37.7% — that is pricing power on a backlog the customers cannot substitute away from. Operating cash flow more than doubled to $766.8M (+153% YoY). Capex guidance for full-year 2026 was raised to $425-525M, attributed to “capacity expansion across the business.” The Americas segment grew 53.1%; Asia Pacific 14.9%; EMEA contracted 20.3% — the only weak spot, and a small enough region of the mix to barely register in the consolidated print. The $15B backlog disclosed in the 10-K was a floor.

Intel confirmed nothing and disclosed something far more important. Revenue beat consensus at $13.6B (+7% YoY). DCAI server ASPs grew 27% with management explicitly describing demand as exceeding their available product supply through the rest of 2026. Intel Foundry posted a $2.4B operating loss but Intel 18A is shipping in product. CCG client volumes were down 13% with ASPs up 16% — supply-constrained, not demand-constrained. Cost discipline showed through: total R&D and MG&A down 8% YoY.

That was the headline. The trapdoor is in the risk-factor language:

“If we are unable to secure sufficient committed demand for Intel 14A through product design wins with potential significant external customers and our Intel products roadmap, we face the prospect that it will not be economical to develop and manufacture Intel 14A and successor leading-edge nodes on a go-forward basis. In such event, we may pause or discontinue our pursuit of Intel 14A.”

This is the first time Intel has formally disclosed that it may exit leading-edge semiconductor manufacturing. The conditional is “external customers and our Intel products roadmap” — which means design wins from Apple, Qualcomm, Nvidia, AMD, the hyperscaler ASIC programs. If those don’t materialise in volume, Intel 14A is paused, Intel becomes a fabless designer at TSMC for everything beyond 18A-P, and the integrated-device-manufacturer premium that has supported the Intel multiple for decades compresses.

Read in this context, the $14.2B Apollo Fab 34 buyout earlier in April makes much more sense. Intel is consolidating ownership of the assets they have before they decide what to do with the assets they don’t yet have. The market, looking at +27% server ASPs and supply-constrained DCAI, may be missing the bigger story underneath: this could be the last fully-vertically-integrated Intel quarter for some time.

Intel may exit leading-edge node manufacturing

Pulling the Intel disclosure into its own thread because the implications are larger than a single earnings reaction.

If Intel pauses 14A, the structural read-across is straightforward. TSMC becomes the only player on leading-edge logic for the next decade — Samsung’s 2nm program is real but unproven, and the only possible disruptor was Intel. With Intel sidelined, TSMC’s pricing power is unconstrained for the customers it actually serves: Apple, Nvidia, AMD, the hyperscalers. That should support TSM’s premium multiple even as the broader semi cycle wobbles.

For AMD, an Intel exit from leading-edge nodes is bullish but with a caveat. The bullish read: Intel’s product competitiveness in the server market caps out at 18A-P-equivalent designs by ~2030. AMD on TSMC continues to skip ahead. The caveat: Intel becoming a TSMC customer eats into the same wafer allocation pool. AMD has the longer relationship and more established design cadence at TSMC, but a credit-rated customer with $13B of quarterly revenue will get capacity.

For Intel itself, the multiple compression scenario is what to model. Today’s price reflects a hybrid valuation — part products business, part foundry option value. If 14A is paused and the foundry option is structurally written down, the stock should re-rate toward the fabless-multiple analogue (call it 12-14× forward earnings) rather than the IDM multiple (~18-22× before the discount for Foundry losses). This is the bear case the disclosure has now made explicit. The market has not fully digested this yet because the disclosure was conditional and embedded in MD&A risk-factor language rather than in the press release.

Tesla joins the hyperscaler capex club

Tesla’s Q1 was, by automotive standards, a normal quarter — revenue $22.39B (+15.7% YoY), net income $477M (+17%), auto gross margin recovering to 21.1% from 16.2%, energy storage margin expanding to 39.5%. The company has $44.7B in cash, generated $3.94B of operating cash flow in the quarter, and 408k vehicles produced.

The non-routine number is the 2026 capex guide:

“We currently expect our capital expenditures to be in excess of $25 billion in 2026, driven by our AI initiatives, including investments in compute infrastructure and data centers, the expansion and ramp of our manufacturing and R&D production lines and facilities.”

Twenty-five billion dollars of capex with explicit attribution to AI initiatives, compute infrastructure, and data centers. Tesla just joined the hyperscaler capex club — a category that until this disclosure included only Microsoft, Google, Meta, Amazon, Oracle, and CoreWeave. The Cortex training cluster at Gigafactory Texas is being expanded; semiconductor fabrication is being brought in-house; an AI hardware company acquisition was signed in April (terms XBRL-masked but paid almost entirely in TSLA stock + equity awards with milestones); and Tesla converted its preferred xAI position into SpaceX common stock under significant-influence accounting.

The auto-manufacturer narrative is no longer sufficient to model the company. A reasonable framing for the next twelve months: Tesla’s automotive segment funds an AI infrastructure platform business that will be evaluated on its own merits by 2027.

For watchlist read-throughs: NVDA picks up another tier-1 customer at scale. VRT likely sees additional cooling/power infra orders as the data center capex lands. The MRVL custom-silicon business may not benefit directly given Tesla’s preference for in-house design — though the AI hardware acquisition could change that calculus.

CoreWeave taps high-yield again — this time at a premium

Issue 01 covered the original CoreWeave triple ($4B convert + $1.75B senior + $1B Jane Street equity over 14-15 April). This week added a fourth tap: an additional $1B in 9.75% Senior Notes due 2031, completed 21 April, issued at 102.000% of par.

The price detail matters. The original $1.75B priced at par on 14 April. The re-tap one week later priced at a slight premium. That tells you the secondary market for this paper firmed up in the week between. High-yield investors would not pay 102 for a re-tap of paper they could buy in the open market at 100. The pricing improvement is modest in absolute terms (~50bps of yield compression) but directionally meaningful: CoreWeave’s cost of incremental high-yield is going down, not up, even as they keep issuing.

Total issuance from CoreWeave across the three-week sprint:

$7.75 billion of capital across 8 days. Use of proceeds language on the latest tap includes “repayment of outstanding indebtedness” — they’re refinancing existing debt while also adding net new capital. This is not a desperation raise; it is a company taking advantage of demand for its paper while the demand exists.

The Intel CAO who left the day after earnings

A small but notable filing on 24 April. Scott Gawel, Intel’s Corporate Vice President and Chief Accounting Officer, resigned effective immediately, “to pursue another career opportunity.” CFO David Zinsner assumed the principal accounting officer role in addition to his existing CFO duties.

CAO resignations the day after a quarterly earnings release are categorically suspicious. The standard playbook: CAO signs off on the quarter, the quarter releases, then the resignation follows if there were disagreements about how items were accounted for. The boilerplate “to pursue another career opportunity” is the language used regardless of actual cause.

This may be nothing — a planned departure timed not to disrupt the quarter close. It may also be a tail signal that the next 10-Q will disclose a control weakness or restatement. The concentration of principal accounting officer + CFO into a single executive at Intel — at a moment when Intel just took on a $6.5B term loan to buy out Apollo from Fab 34 and disclosed the conditional 14A pause — is itself a material risk factor.

Watch the next 10-Q. If a control-weakness disclosure follows in 90 days, the alert escalates. For now: yellow flag.

The week ahead — second wave of earnings

Last week tested VRT, DLR, INTC, and TSLA. This week tests the second cohort:

The Big Tech triple-header on 30 April is the moment of the week. Microsoft’s $155B in not-yet-commenced operating leases, Meta’s $70-72B 2026 capex pace, and Google’s $113B in short-term purchase commitments will be tested against actual quarterly print. If any of the three guides 2026 capex down — even slightly — the entire AI infrastructure complex will reprice.

Watchlist alerts

None at the threshold this week. The earnings wave has not yet generated enough cross-source convergence on off-watchlist tickers to clear the 3-source / 30-day bar. Expect that to change as the second cohort reports.

Anomalies

The market was closed Saturday and Sunday; the most recent trading day at filing time was Friday 24 April. Notable anomalies in the week without filing context: minimal — the week’s price action was largely driven by the earnings releases themselves. The dashboard’s anomaly detector will repopulate Tuesday morning with Monday’s close.

—

Source dashboard: http://34.21.182.120:8051 for the live data and signal-by-signal detail.