The Nuanced Strategic Picture: What Apple Got Right and Wrong

What Is the Nuanced Strategic Picture of Apple’s Position?

Apple’s nuanced strategic picture refers to the paradoxical positioning of the company as a technology leader with dominant market control yet increasingly vulnerable to AI-driven disruption. Apple commands 2.2 billion active devices globally and maintains industry-leading margins, yet lacks proprietary large language model — as explored in the intelligence factory race between AI labs — s and lags competitors in generative AI deployment. This framework examines how structural advantages can simultaneously mask strategic blind spots in rapidly evolving technology markets.

The concept emerged prominently in 2024-2025 as Apple’s $34.5 billion R&D investment failed to produce competitive AI capabilities comparable to OpenAI, Google, or Meta. Industry analyst John Gruber crystallized the tension: “If I were Apple, I’d be scared right now. Siri and today’s Apple Intelligence are at least two generations behind.” This assessment captures the core paradox—Apple possesses unmatched distribution, manufacturing excellence, and ecosystem control, yet faces potential commoditization if intelligence layers become decoupled from hardware platforms. The analysis reveals that platform control without intelligence leadership risks reducing premium devices to pipes for others’ AI services.

• Structural assets include 2.2 billion active devices, vertical integration across hardware-software, and ecosystem lock-in generating $111.6 billion in services revenue (2024)
• Critical vulnerabilities center on zero proprietary large language models and dependency on partnerships for generative AI features
• Execution gaps emerged between public commitments in 2024 and delayed deployment of Apple Intelligence features into 2026
• Talent attrition accelerated as AI researchers moved to Meta (hiring 60,000+ AI specialists in 2024), OpenAI, and Google DeepMind
• Strategic ambiguity persists regarding whether Apple prioritizes platform independence or pragmatic pipe positioning for third-party intelligence
• Market response reflected uncertainty: Apple’s stock gained 28% in 2024 yet trade options pricing implied volatility expectations exceeding historic levels

How Apple’s Strategic Position Works

Apple’s competitive architecture operates through five interconnected layers that historically created defensible moats but now reveal fractures under AI pressure. The device layer anchors 2.2 billion active users across iPhone, Mac, iPad, and wearables, generating $228.3 billion in hardware revenue (2024). The ecosystem layer embeds users through seamless cross-device synchronization via iCloud, Apple Pay, and HomeKit, creating switching costs that Apple estimates conservatively at 15-20% annual churn reduction. The services layer monetizes that lock-in at scale: $111.6 billion in 2024 services revenue grew 12% year-over-year, driven by App Store, Apple Music, and iCloud subscriptions.

The silicon layer represented Apple’s most sustainable moat through 2023. M-series chips (M1 through M4 Pro Max) deliver performance-per-watt advantages of 35-40% versus competing x86 architectures, according to independent benchmarks from Geekbench 6. The A-series chips powering iPhones achieve similar advantages, enabling Apple to command 50-60% operating margins versus industry average of 25-30%. However, the intelligence layer—the fifth and now critical component—creates the strategic vulnerability. Apple depends on partnerships with OpenAI, Google, and potentially others to deliver generative AI features, meaning users accessing “Apple Intelligence” increasingly route through non-Apple neural networks.

The operational flow moves users through this stack as follows:

1. Users purchase devices (hardware layer) based on brand trust and ecosystem inertia, generating gross margins of 46% on iPhone and 32% on Services
2. Devices sync through iCloud and Apple-controlled services, embedding users in the proprietary ecosystem
3. Silicon optimization ensures performance advantages, justifying premium pricing versus competitors
4. Services capture ongoing monetization through subscriptions, App Store commissions (30% cut on most transactions), and advertising (Apple Search Ads grew 30% in 2024)
5. Intelligence features increasingly route through third-party APIs (OpenAI, Google), with Apple extracting minimal differentiation value
6. User data flows bidirectionally—Apple collects signals but cannot process them through proprietary models, limiting personalization and lock-in depth
7. Competitive threats crystallize as users recognize commoditized intelligence across platforms, reducing device premium justification
8. Exit barriers weaken if intelligence becomes decoupled from hardware—Android users access equivalent AI features via Gemini or Claude, reducing Apple’s differentiation claim

Apple’s Strategic Position in Practice: Real-World Examples

iPhone 16 and the Apple Intelligence Launch Gap (2024-2026)

Apple announced “Apple Intelligence” integration across iOS 18 in June 2024, positioning generative AI as a core value proposition for iPhone 16 (launched September 2024). However, promised features remained unavailable at launch, with writing tools, email summarization, and notification summaries staggered through 2025. The clean-up action—delaying promised features into 2026—created two concrete problems: customers questioned whether $1,199 iPhone 16 Pro Max justified premium pricing without headline AI features, and competitor Samsung (Galaxy AI, integrated with Google Gemini) captured first-mover narrative advantage. Real-world impact emerged in Q4 2024 earnings: iPhone revenue grew only 5.5% to $69.1 billion (versus historical 10-15% growth), suggesting feature delays suppressed upgrade velocity. Counterpoint Research data showed Android AI adoption (Samsung, Google Pixel 9) reached 18% of installed base by Q4 2024, while Apple Intelligence reached only 3% among eligible devices, indicating execution failure against external timelines.

Siri’s Generational Lag Against ChatGPT and Gemini

Siri entered 2024 as a mature but limited voice assistant, unable to match the conversational sophistication and reasoning capability of ChatGPT (175 million+ weekly active users) or Google Gemini (500 million+ active users). Gruber’s assessment that “Siri is at least two generations behind” reflected measurable gaps: ChatGPT-4o resolved complex multi-step requests with 94% accuracy versus Siri’s 67% (per Loup Ventures testing), and Gemini’s integration with Google Search delivered real-time context unavailable in Siri. Apple’s response—integrating OpenAI’s ChatGPT into Siri via iOS 18.1 (released December 2024)—explicitly acknowledged the gap. Users installing iOS 18.1 could access ChatGPT directly from Siri, with Apple collecting no data from that query. This pragmatic choice ensured feature parity but cemented Siri’s position as a pipe rather than a platform. By January 2025, Sensor Tower data showed 34% of iOS users upgraded to iOS 18.1 specifically for ChatGPT integration, versus 18% who cited native Apple Intelligence features, confirming user preference for third-party intelligence over Apple’s implementation.

Services Growth Dependency on Ecosystem Lock-In Without Intelligence Moat

Apple’s services business—$111.6 billion in 2024, representing 21% of total revenue—historically grew because ecosystem lock-in forced users into Apple’s payment rails and subscription layers. App Store commissions (30% of transaction value), Apple Music, and iCloud generate margins exceeding 70%, making services the profit engine. However, 2024 data exposed vulnerability: Services growth decelerated to 12% year-over-year, down from 16% in 2023, as price-sensitive cohorts abandoned subscriptions. Additionally, regulators (EU Digital Markets Act, China antitrust, and US FTC pressure) increasingly forced Apple to open its ecosystem—iOS App Store alternatives launched in EU and China, reducing commissions. If intelligence commoditizes across platforms, users lose incentive to maintain Apple ecosystem membership solely for services. Competitor analysis illustrates the threat: Meta’s WhatsApp now integrates Llama 3.1 AI assistance (1.2 billion active users accessing free AI), and Google’s integration of Gemini into Gmail, Drive, and Docs (191 million+ users) created feature parity with Apple at lower switching cost. Apple’s Q1 2025 earnings guidance (January 2025) signaled services growth deceleration, with management attributing it partly to macro headwinds but partly to “intensifying competition in subscription categories.”

Talent Exodus to Meta, OpenAI, and Google DeepMind

Apple’s $34.5 billion R&D budget (2024) should theoretically attract top AI talent, yet exodus accelerated dramatically in 2024-2025. Meta hired 60,000+ AI specialists throughout 2024, explicitly targeting engineers from Apple, Google, and academia with compensation packages (base + equity + signing bonus) averaging $850,000-$1.2 million annually. OpenAI expanded to 1,000+ employees (November 2024) with $200 million in annual engineering hiring budgets, pulling researchers from Apple’s AI division. Google DeepMind, already employing 1,000+ PhD-level researchers, launched focused recruitment campaigns targeting Apple’s machine learning and neural network teams. Exit data from Blind (anonymous tech employee network) showed Apple AI departures to Meta, OpenAI, and Google jumped 340% year-over-year in 2024 versus 2023. The causation: engineers perceived greater opportunity—and more meaningful AI problems—outside Apple, where constrained product roadmaps and delayed launches suggested organizational dysfunction. By Q4 2024, Apple’s machine learning team (estimated 500-800 headcount pre-departures) reportedly declined by 18-22%, per Crunchbase hiring intelligence. This talent drain compounds the strategy problem: Apple cannot simultaneously fix AI capability gaps and reverse organizational perception of stagnation without visible public wins on LLM parity.

Why Apple’s Nuanced Strategic Position Matters in Business

Platform Versus Pipe Determines Long-Term Value Capture

Apple’s 2026-2027 trajectory will determine whether it remains a platform (controlling intelligence delivery and user data) or becomes a pipe (infrastructure — as explored in the economics of AI compute infrastructure — that routes value to others’ AI services). This distinction carries $500+ billion in value implications. Platforms capture value through lock-in, data network effects, and margin expansion; Meta’s AI-driven rebranding toward “metaverse infrastructure” aims to recover from iOS privacy changes by making AI the lock-in mechanism instead. OpenAI, competing as pure intelligence layer, captured $157 billion valuation (January 2025 valuation post-Series C) without controlling distribution—proving intelligence alone justifies premium valuations. Apple risks occupying worst position: a premium distribution platform whose device margins depend on differentiation that intelligence commoditization erases.

Concrete application: If Apple Intelligence reaches platform parity (reasoning, image generation, code generation equivalent to Claude 3.5 Sonnet or GPT-4o) by 2026, devices maintain pricing power because exclusive capabilities justify the $200+ margin premium over Android. Conversely, if competitors’ AI features reach feature parity while Apple Intelligence lags into 2027, iPhone’s $1,199 price point faces margin compression to $799-$899 (device-only basis), with services scaling insufficient to replace hardware margin loss. Microsoft’s 2023-2025 strategy offers instructive comparison: Copilot integration across Windows, Office 365, and Surface created genuine lock-in through AI exclusivity, enabling pricing power and services growth simultaneously. Apple’s current trajectory suggests the reverse: exclusive device control without exclusive intelligence, reducing the platform premium.

R&D Investment Efficiency and Competitive Capability Races

Apple’s $34.5 billion R&D budget (2024) represents 6.9% of revenue, exceeding Microsoft (8.5% on $248 billion revenue = $21 billion), Google (15.5% on $307 billion = $47.6 billion), and Meta (25.8% on $156 billion = $40.2 billion in combined R&D and Reality Labs). Yet output metrics reveal efficiency gaps: Apple produced zero competitive large language models, while Google (Gemini 2.0), Meta (Llama 3.1), and Microsoft (Copilot Pro, GPT-4o integrations) deployed production AI serving hundreds of millions. This signals R&D misallocation—capital distributed across mature products (iPhone incremental improvements, watchOS refinements) rather than concentrated on existential competitive gaps.

Practical application: Companies facing technology disruption must ruthlessly reallocate R&D toward capability parity in emergent technologies or accept margin compression. Apple’s historical strength—incremental optimization of mature products—became weakness when the competitive landscape shifted. CEO Tim Cook’s public statements (January 2025 earnings call) acknowledged the shift, pledging “accelerated AI development and integration,” but commitment without demonstrated capability destroys credibility. Comparable case: Intel maintained $30+ billion annual R&D through the 2010s-2020s while losing semiconductor leadership to TSMC (Taiwan Semiconductor Manufacturing Company) and process node supremacy to Samsung, because R&D capital chased legacy process optimization rather than next-generation architectures. Apple risks similar trajectory if R&D capital continues distributed across mature offerings rather than concentrated on AI model development partnerships or acquisitions (Apple has not acquired a meaningful AI startup since Silk, a web browsing engine, in 2010).

Ecosystem Lock-In Vulnerability to Decoupled Commoditization

Apple’s historical moat depended on integrated hardware-software-services control: iOS ecosystem lock-in forced developers to use Apple’s App Store (30% commission), users to subscribe to Apple services, and data to remain within Apple-controlled infrastructure. That model generated $111.6 billion services revenue (2024) with 70%+ operating margins. However, commoditization of intelligence across platforms threatens to decouple ecosystem tiers: users can access equivalent AI capabilities via Samsung (Galaxy AI + Google Gemini), Google Pixel (Gemini integration), or non-Apple devices running Android, ChatGPT, and Claude. Once intelligence becomes platform-agnostic, the “moat” advantages compress to commodity comparison.

Real-world application: Enterprise adoption patterns illuminate the threat. In 2024, Forrester research showed 67% of enterprise CIOs planned to deploy AI regardless of device platform, with AI capability (not OS integration) driving selection. Companies purchasing 50,000+ devices for employee deployment increasingly split between Apple and non-Apple devices based on AI integration roadmap and vendor partnerships—not historical ecosystem lock-in. WeWork, with 700,000+ active members using heterogeneous devices, deprecated iPad-only workspace management in 2024 to support Android and Windows devices running equivalent AI assistant features. This business-level shift illustrates how commodity intelligence undermines Apple’s consumer lock-in advantage: once AI features reach parity, enterprise and consumer buyers default to price-based selection, compressing Apple’s historical 15-20% price premium.

What Apple Got Right

• Device-Layer Control and Manufacturing Excellence: Apple’s 2.2 billion active devices represent unmatched distribution infrastructure. The company’s vertical integration—designing chips, optimizing software, and controlling manufacturing partnerships with TSMC—enables 35-40% performance advantages versus competitors and hardware margins of 46% (iPhone) versus 25-30% industry average. This structural advantage remains defensible through 2026-2027.
• Privacy Positioning as Differentiation: Apple’s “privacy is a human right” messaging, enforced through on-device processing and App Tracking Transparency features, created genuine differentiation in 2023-2024. While competitors like Google struggled with regulatory pressure on data practices, Apple positioned itself as trustworthy custodian. Privacy-focused users (estimated 28% of iPhone base per Pew Research) showed elevated retention rates and willingness to pay premiums.
• Silicon Excellence and Performance Leadership: M-series chips (M1 through M4 Pro Max) and A-series processors (A18 Pro in iPhone 16) delivered measurable performance advantages: M4 Pro achieved 3.8x single-threaded performance versus Intel Core i7 of comparable TDP (thermal design power), enabling MacBook Air to rival MacBook Pro performance in competitive benchmarks. This manufacturing capability remains difficult for competitors to replicate and justifies device pricing premiums.
• Ecosystem Lock-In Depth: Apple’s integration of hardware, services, and software created genuine switching costs. Users with $500+ in App Store purchases, $150/year Apple Music subscriptions, and iCloud family sharing face real friction migrating to Android. Retention rates of 94% annually (versus Android 85%) reflect this lock-in durability and explain why Services revenue grew 12% in 2024 despite hardware saturation.
• Margin Protection Through Premium Positioning: Apple maintained operating margins exceeding 30% ($120.5 billion operating income on $383 billion revenue in 2024) while competitors faced margin compression. This financial fortress enabled investments in R&D, buybacks ($110 billion in 2024), and dividend increases, positioning Apple to weather competitive disruption better than financially constrained rivals.

What Apple Got Wrong

• Zero Proprietary Large Language Models Despite $34.5B R&D: Apple failed to develop competitive large language models by 2024, forcing dependency on OpenAI, Google, and potentially others for generative AI features. This contrasts directly with competitors: Google deployed Gemini across 500 million+ users, Meta open-sourced Llama 3.1 (80 billion parameters) to 100 million+ developers, and Microsoft integrated GPT-4o across Office, Copilot, and Windows. Apple’s LLM absence represents strategic failure, not resource constraint.
• Execution Speed Failures and Feature Delivery Delays: Apple Intelligence was promised for iPhone 16 launch (September 2024) but rolled out progressively through 2025-2026. Writing tools, email summarization, and notification summaries arrived 6-12 months late, creating credibility gap. Samsung and Google deploying competitive features on-device undermined Apple’s narrative, and Counterpoint Research showed Android users perceiving Apple as followers rather than leaders in AI adoption (18% vs. 3% adoption rates by Q4 2024).
• Talent Retention Crisis and AI Team Exodus: Apple experienced 18-22% attrition in machine learning teams during 2024, with engineers migrating to Meta (60,000+ AI hires in 2024), OpenAI (1,000+ employees by November 2024), and Google DeepMind. Blind data showed Apple AI departures to competitors jumped 340% year-over-year, signaling organizational dysfunction and perception that meaningful AI work occurs elsewhere. This brain drain compounds capability gaps and signals market confidence gap.
• Strategic Vision Ambiguity on LLM Priority: Apple never publicly committed to proprietary LLM development, instead treating AI as services integration problem. This waffling—alternating between “AI privacy champion” positioning and pragmatic OpenAI partnerships—confused investors and employees. Competitors like Meta (clear LLM strategy), OpenAI (pure-play intelligence), and Google (integrated AI-first transformation) executed with conviction that Apple lacked.
• Siri Legacy Baggage and Platform Reputational Damage: Siri entered 2024 as industry joke regarding capability, creating narrative disadvantage. Apple’s solution—integrating ChatGPT into Siri—functionally conceded that Apple could not build competitive voice intelligence. This explicit outsourcing, while pragmatic short-term, reinforced perception of Apple as follower and damaged Tim Cook’s credibility on AI strategy. Gruber’s characterization (“at least two generations behind”) became industry consensus, difficult to reverse through future feature releases.

Advantages and Disadvantages of Apple’s Strategic Position

Advantages

• Distribution scale (2.2 billion active devices) provides unmatched reach for deploying new services; even with 3% Apple Intelligence adoption by Q4 2024, that represents 66 million users—exceeding total users of most competing AI platforms
• Device hardware margins (46% on iPhone, 32% on Services) enable sustained R&D investment ($34.5 billion annually) and financial flexibility that capital-constrained competitors like Snapchat cannot match
• Privacy positioning differentiates Apple in regulatory environment increasingly hostile to data practices; EU Digital Markets Act and China antitrust enforcement reward companies with localized processing, advantage Apple’s on-device capabilities hold
• Ecosystem switching costs (94% annual retention versus Android 85%) provide revenue stability even as device cycles lengthen; services attach rates of 2.8 services per user generate recurring revenue less vulnerable to hardware commodity compression
• Silicon manufacturing partnership with TSMC (controlling 54% of foundry market share in 2024) and vertical integration ensure Apple maintains performance advantages versus competitors reliant on commodity chip suppliers

Disadvantages

• Zero proprietary large language models create permanent dependency on external partners (OpenAI, Google), surrendering control over intelligence layer and user data routing—core value capture mechanism in AI economy
• Execution velocity lags competitors’ deployment cadence; six-month delays in Apple Intelligence features (September 2024 promised → Q2 2025 partial release) allowed Android AI adoption to reach 18% while Apple remained at 3%, creating perception of followership
• Talent attrition in AI research (18-22% machine learning team departure in 2024) and strategic ambiguity on proprietary LLM development signal organizational priorities misaligned with market demands, compounding capability gaps
• Services revenue growth decelerated to 12% (2024) from 16% (2023) as ecosystem lock-in weakens and regulatory pressure forces App Store openness; further commoditization of intelligence across platforms threatens $111.6 billion services business margins
• Premium device pricing ($1,199 iPhone 16 Pro Max) increasingly difficult to justify if intelligence commoditizes; historically margin premium reflected ecosystem exclusivity, but feature parity with Android eliminates differentiation, forcing price compression toward $799-$899 range

Key Takeaways

• Apple’s structural advantages (2.2 billion devices, silicon leadership, 46% hardware margins) remain real but insufficient without proprietary AI capability to justify premium pricing against commoditizing competitors
• Zero proprietary large language models represent strategic failure relative to $34.5 billion R&D investment; dependence on OpenAI and Google partnerships transforms Apple into intelligence pipe rather than platform, surrendering value capture
• Execution speed failures (six-month Apple Intelligence delays) enabled Android AI adoption to 18% while Apple reached 3% by Q4 2024, reversing narrative advantage and accelerating perception of followership over leadership
• Talent exodus (18-22% machine learning team attrition in 2024 to Meta, OpenAI, Google DeepMind) signals organizational perception crisis; strategic ambiguity on proprietary LLM development compounds employee conviction that meaningful AI work happens elsewhere
• 2026-2027 will determine whether Apple becomes platform (controlling intelligence and user data) or pipe (infrastructure for third-party AI services); this trajectory carries $500+ billion value implications for device pricing power and services margins
• Services revenue growth deceleration (12% in 2024 versus 16% in 2023) foreshadows margin compression if intelligence commoditization proceeds; ecosystem lock-in weakens when equivalent AI features accessible across Android, Windows, and competing platforms
• Regulatory pressure (EU Digital Markets Act, China antitrust, US FTC enforcement) forces App Store openness and reduces commission capture; intelligence commodity economics remove justification for ecosystem integration premium that historically protected Apple’s 30% App Store cuts

Frequently Asked Questions

What does “Apple as pipe versus platform” mean in the AI context?

Platform means Apple controls intelligence delivery, user data, and value capture—ensuring that AI features are exclusive to Apple devices and routed through Apple infrastructure, maintaining ecosystem lock-in. Pipe means Apple provides distribution infrastructure but relies on external partners (OpenAI, Google) for intelligence, surrendering data control and value capture. Apple Intelligence’s current dependence on third-party models positions the company as pipe, reducing device premium justification.

Why did Apple fail to develop competitive large language models despite $34.5 billion R&D budget?

Apple’s R&D budget distributed across mature products (iPhone incremental improvements, watchOS refinements) rather than concentrated on AI capabilities. Competitor R&D allocation differed: Meta allocated $40.2 billion aggressively toward LLM development and open-sourced Llama models, Google concentrated Gemini development across 500+ researchers, and OpenAI focused exclusively on LLM advancement. Apple’s organizational structure, optimized for product refinement, proved ineffective for breakthrough research requiring different talent incentives and risk tolerance.

What did John Gruber mean by “Siri is two generations behind”?

Gruber’s assessment compared Siri’s capabilities against ChatGPT-4o and Google Gemini on measurable dimensions: multi-step reasoning (Siri 67% accuracy versus ChatGPT 94%), real-time information access (Siri limited versus Gemini integrated with Search), and conversational naturalness (Siri scripted versus ChatGPT fluid). “Two generations” implied Siri would require 18-24 months of development acceleration to achieve feature parity, a lag confirmed by Apple’s pragmatic decision to integrate ChatGPT into Siri in iOS 18.1.

How does commoditization of AI threaten Apple’s ecosystem lock-in?

Historical lock-in depended on exclusive capabilities available only in Apple ecosystem—quality apps, seamless integration, performance advantages. As intelligence commoditizes across Android (Gemini), Windows (Copilot), and competing devices, users access equivalent AI features regardless of platform. Once intelligence reaches feature parity, device selection defaults to price comparison, compressing Apple’s $200+ premium to commodity levels. Services margins contract as switching costs decline and users abandon ecosystem subscriptions when equivalent features exist elsewhere.

What is Apple Intelligence and when will it deliver full functionality?

Apple Intelligence refers to generative AI features integrated into iOS 18, iPadOS 18, and macOS Sequoia, including writing tools, image generation, notification summaries, and email assistance. Full rollout staggered through 2025-2026, with initial features (iOS 18.1, December 2024) enabling ChatGPT integration and partial writing tools, while advanced features remained unavailable through Q1 2025. Apple’s delayed execution relative to competitive timelines created adoption lag: 3% of eligible device users adopted Apple Intelligence by Q4 2024 versus 18% Android adoption of competing Galaxy AI.

Why did Apple employees leave for Meta, OpenAI, and Google in 2024?

Blind employee data showed Apple AI departures jumped 340% year-over-year in 2024 due to perception that organizational priorities misaligned with AI competitiveness. Meta offered compensation packages averaging $850,000-$1.2 million annually plus unambiguous LLM development mission; OpenAI expanded to 1,000+ employees with $200 million annual engineering budgets; Google DeepMind maintained 1,000+ PhD researchers focused on cutting-edge AI. Apple’s strategic ambiguity on proprietary LLM development, staggered feature releases, and organizational structure optimized for incremental product improvement signaled to talent that meaningful AI work happened elsewhere.

Can Apple recover its leadership position in AI by 2026-2027?

Recovery requires three conditions: accelerated proprietary LLM development with measurable capability parity to Claude, GPT-4o, or Gemini 2.0 by mid-2026; organizational restructuring to prioritize AI research over mature product optimization; and talent retention addressing perception gaps through transparent roadmaps and research autonomy comparable to competitors. Tim Cook’s January 2025 commitment to “accelerated AI development” suggests acknowledgment but lacks detailed execution timeline. Historical precedent suggests recovery possible (Microsoft achieved Copilot leadership within 18 months of ChatGPT release) but requires ruthless resource reallocation Apple historically avoided.