Some Apple AI servers are reportedly sitting unused on warehouse shelves, due to low Apple Intelligence usage

Apple is reportedly in advanced talks to host the next-generation Siri models inside Google data centers because its Private Cloud Compute is underpowered, difficult to update and only about 10% utilized, leaving manufactured servers idle. The Private Cloud uses modified M2 Ultra-class chips that are said to lack capacity to run frontier models like Google’s Gemini, prompting Apple to lean on Google’s proven LLM infrastructure while still maintaining Apple privacy standards; the company may pursue heavier in-house investment over a longer horizon but has repeatedly stalled on unifying fragmented internal cloud stacks.

Analysis

Market structure: Google (GOOGL/GOOG) is the clear near-term winner — hosting Siri shifts high-margin services revenue and recurring cloud spend toward Google Cloud and raises demand for high-end accelerators (benefit NVDA, AMZN to a lesser extent). Apple (AAPL) is a near-term loser from headline optics (wasted Private Cloud Compute capex, potential services margin pressure) but not existentially impaired; expect modest re-rating risk of 3–8% if investor attention focuses on wasted CapEx or slower Siri rollouts over 3–12 months. Higher demand for GPUs/TPUs tightens supply; expect cloud spot rates for accelerator instances to rise 10–30% vs. current tails if other hyperscalers follow.

Risk assessment: Tail risks include privacy/regulatory pushback (FTC/Europe) against outsourcing Siri model hosting to Google, or a high-profile data incident causing user trust erosion — both could knock 5–15% off AAPL services expectations within 6–12 months. Operational risks for Google (outage, model performance gaps, contract disputes) could blunt upside; monitor contract scope and SLAs. Catalysts: Apple WWDC (June 2026), Google Cloud earnings/guide (next 1–2 quarters), NVDA capacity guidance; any of these could accelerate repricing.

Trade implications: Favor overweight Google Cloud exposure (GOOGL) and AI infra names (NVDA, AMZN) over Apple hardware-exposed risk; implement option structures to express skewed upside. Use pair trades (long GOOGL, short AAPL) to isolate cloud-hosting thesis from broad tech beta; target 1–3% portfolio exposure per leg and adjust on 5% moves. Entry window: initiate within 2–6 weeks, scale on positive Google Cloud commentary or AAPL admissions; exit or re-evaluate around WWDC (by June 2026) and next earnings cycle (by Oct 2026).

Contrarian angle: The market may overstate permanent damage to AAPL — outsourcing could be a pragmatic cost save that boosts services margin long-term if Apple negotiates favorable economics; downside may be capped at single-digit percent absent regulatory action. Conversely, investors underestimate contract lock-in risk: Google could extract pricing leverage over time, pressuring Apple margins beyond the initial hosting fees. Historical parallel: early cloud outsourcing deals often led to later vendor lock-in and price renegotiations (AWS/Netflix), implying durable revenue for the cloud provider rather than one-time gains.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in GOOGL within the next 2 weeks to capture Google Cloud upside; add another 1% on a pullback >5% or after positive Cloud guidance (target 6–12 month price appreciation of 5–15%).
• Allocate 1–2% to NVDA via 3–6 month call spreads (buy 20–30% OTM calls, sell 40–60% OTM calls) to play increased demand for accelerators while limiting premium outlay; roll or realize on NVDA capacity commentary or quarterly results.
• Reduce AAPL exposure by 1–2% or hedge with 3-month 5% OTM puts sized to cover that delta (buy puts covering 1–2% portfolio) ahead of WWDC (by June 2026) to protect versus service-capex downside or negative Siri reception.
• Execute a pair trade: long 1.5% GOOGL and short 1.0% AAPL to isolate cloud-hosting arbitrage; rebalance if the spread moves >7% (trim winner by half, add to loser by half) and reassess around Apple’s next earnings and WWDC.
• Monitor three specific triggers in the next 30–90 days before adding: (1) Apple disclosure on Siri hosting/contract terms, (2) Google Cloud revenue/guide vs. consensus, and (3) NVDA supply/capacity updates — if two of three are positive, increase AI infra exposure by another 0.5–1.0%.