Apple Inc.
Last reviewed
May 1, 2026
Sources
47 citations
Review status
Source-backed
Revision
v1 ยท 4,419 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 1, 2026
Sources
47 citations
Review status
Source-backed
Revision
v1 ยท 4,419 words
Add missing citations, update stale details, or suggest a clearer explanation.
Apple Inc. is an American multinational technology company headquartered at Apple Park in Cupertino, California. It was founded as Apple Computer Company on April 1, 1976 by Steve Jobs, Steve Wozniak, and Ronald Wayne; incorporated as Apple Computer, Inc. on January 3, 1977; and renamed Apple Inc. on January 9, 2007 to reflect its move beyond the personal computer business [1][2][3]. Apple is the world's most valuable publicly traded company, with a market capitalization above $4 trillion in early 2026, fiscal-2024 revenue of $391 billion, fiscal-2025 revenue of $416 billion, and roughly 165,000 to 166,000 full-time employees worldwide [4][5]. The company has been led by CEO Tim Cook since August 24, 2011, with Arthur Levinson serving as chairman of the board; Cook announced in April 2026 that he would step down on September 1, 2026 in favour of hardware engineering chief John Ternus [5][6].
Within the machine learning and AI industry, Apple is best understood as a vertically integrated platform company that bets on on-device inference rather than centralised cloud models. It designs its own AI accelerator (the Apple Neural Engine, shipped on every flagship Apple SoC since the A11 Bionic in 2017), its own ML runtime (Core ML, introduced at WWDC 2017), its own foundation models (the on-device 3-billion-parameter and server PT-MoE models that power Apple Intelligence since 2024), and its own server hardware (Private Cloud Compute, running on Apple silicon) [7][8][9]. The combination of Apple's hardware install base (over two billion active devices) and its long-running privacy-first positioning has made it one of the largest deployers of consumer-facing AI in the world, even though it arrived publicly in the generative-AI race well after OpenAI, Google, and Anthropic.
Apple's current product lineup spans the Mac (Apple Silicon since 2020), the iPhone (~51% of fiscal-2024 revenue), the iPad, the Apple Watch, AirPods, the Apple TV set-top box, the HomePod and HomePod mini, and the Apple Vision Pro mixed-reality headset (launched February 2024) [4][10]. Services contributed $96.2 billion in fiscal-2024 revenue and remain the company's fastest-growing segment [4]. R&D spending reached approximately $31.4 billion in fiscal 2024, much of it directed at silicon, AI, and the Vision Pro line [4].
| Metric | Value |
|---|---|
| Founded | April 1, 1976 [1][2] |
| Founders | Steve Jobs, Steve Wozniak, Ronald Wayne [1][2] |
| Incorporated | January 3, 1977 (Apple Computer, Inc.) [2] |
| Renamed to Apple Inc. | January 9, 2007 [3] |
| Headquarters | Apple Park, Cupertino, California (since April 2017) [11] |
| CEO | Tim Cook (since August 24, 2011; departure announced for September 1, 2026) [5][6] |
| Chairman | Arthur Levinson [12] |
| SVP Machine Learning and AI Strategy | John Giannandrea (joined April 2018) [13][14] |
| FY2024 revenue | $391.0 billion [4] |
| FY2024 R&D spending | ~$31.4 billion [4] |
| FY2025 revenue | ~$416 billion [5] |
| Employees | ~165,000-166,000 [5] |
| Market capitalization (early 2026) | >$4 trillion [5][15] |
| Active device install base | >2.2 billion (reported FY2024) [4] |
Apple's legal headquarters is at One Apple Park Way in Cupertino, the circular Norman Foster-designed building informally known as the "spaceship" that opened to employees in April 2017 and replaced the older Infinite Loop campus [11]. The Vision Pro launch in 2024 and the Apple Intelligence rollout that began later the same year are the two largest product bets of Cook's late tenure.
Unlike most large-scale AI labs, Apple does not have a single "AI division" with a public org chart. Machine learning work at Apple is split across product groups (Photos, Camera, Siri, Health, Maps, Vision Pro), the Hardware Engineering team that owns Apple Silicon and the Apple Neural Engine, and a central AIML group led by senior vice president John Giannandrea, who joined Apple in April 2018 from Google, where he had spent eight years leading the Machine Intelligence, Research and Search teams [13][14]. Giannandrea was named to the Apple executive team in December 2018; his portfolio includes Core ML, Siri, search, and the broader Apple Intelligence effort.
Other figures relevant to Apple's AI work include Craig Federighi (SVP Software Engineering, who hosts most public AI demos), Johny Srouji (SVP Hardware Technologies, who has run the Apple Silicon program since the A4 in 2010), and on the research side Carlos Guestrin (former Turi co-founder, longtime director of machine learning at Apple), Ruslan Salakhutdinov (Apple's first director of AI research from 2016 to 2020), and Samy Bengio (who joined in 2021 from Google) [16][17]. Apple does not disclose total ML headcount, but reporting puts the number across product groups and the AIML organisation in the low thousands.
Apple's modern AI program dates to the Siri acquisition in 2010 and accelerated sharply after Tim Cook hired Giannandrea from Google in 2018, although the company's involvement in machine learning predates the deep-learning era (Newton handwriting recognition in 1993, iPhoto face detection in 2002, iPhone keyboard statistical correction in 2007).
| Year | Event |
|---|---|
| 2010 | Apple acquires Siri Inc. in April for a reported price above $200 million [18][19]. |
| 2011 | Siri launches as a built-in feature on the iPhone 4S on October 4, the first mass-market AI voice assistant on a smartphone [19][20]. |
| 2014 | Apple acquires Beats Music and Beats Electronics for roughly $3 billion; the Beats team becomes the foundation of Apple Music [21]. |
| 2015 | Apple acquires VocalIQ (UK speech AI) and Perceptio (on-device image classification) to strengthen Siri and Photos [16]. |
| 2016 | Apple acquires Turi, a Seattle ML platform led by Carlos Guestrin, for a reported $200 million; Apple deploys differential privacy in iOS 10 [22][23]. |
| 2017 | Core ML and the Vision and Natural Language frameworks are introduced at WWDC. The A11 Bionic and its first Apple Neural Engine ship in the iPhone 8, 8 Plus, and iPhone X in September [24][25]. |
| 2018 | Apple hires John Giannandrea from Google in April; he is named to the executive team as SVP of ML and AI Strategy in December. Create ML ships at WWDC [13][14]. |
| 2019 | Apple acquires Drive.ai in June after Project Titan layoffs [26]. The A13 Bionic introduces the Deep Fusion computational photography pipeline. |
| 2020 | Apple acquires Xnor.ai, the AI2 spin-out, for a reported $200 million in January [27][28]. The M1 ships in November, beginning the Mac transition off Intel [29]. |
| 2021 | Siri's on-device speech recognition launches with iOS 15. Live Text adds OCR across the camera and Photos. The A15 Bionic doubles ANE throughput again. |
| 2022 | Apple acquires AI Music in February [30]. The Photonic Engine debuts on the iPhone 14 Pro. ml-ane-transformers is released. |
| 2023 | The A17 Pro on the iPhone 15 Pro debuts a 35 TOPS ANE, the first Apple chip large enough to host a multi-billion-parameter foundation model [31]. Apple is reported to be developing an internal LLM project codenamed Ajax [32]. |
| 2024 (June 10) | Apple Intelligence is announced at WWDC, with an OpenAI partnership for ChatGPT handoff. The "Apple Intelligence Foundation Language Models" report follows in late July [7][33][34]. |
| 2024 (October 28) | Apple Intelligence ships in iOS 18.1, iPadOS 18.1, and macOS Sequoia 15.1 in US English, with Writing Tools, Notification Summaries, Genmoji, Image Playground, and the first phase of the new Siri [35]. |
| 2025 (March) | Apple confirms the more personalised, on-screen-aware Siri features are delayed to spring 2026, citing problems with the V1 architecture [36][37]. |
| 2025 (June, WWDC25) | The Foundation Models framework is released to third-party developers [9][38]. |
| 2026 (January 12) | Google and Apple announce a multi-year partnership using a custom 1.2-trillion-parameter Gemini model behind some Apple Intelligence features, including the rebuilt Siri [39][40]. |
| 2026 (April) | Apple announces Tim Cook's transition on September 1, 2026 and confirms John Ternus as the next CEO [6]. |
Apple ships AI through a combination of consumer-facing features (Siri, Photos, Live Text, Visual Look Up, Apple Intelligence) and developer frameworks (Core ML, the Vision and Natural Language frameworks, the Foundation Models framework, Create ML). Almost all of these run on the Apple Neural Engine when the device supports it.
| Product or platform | Year | Description |
|---|---|---|
| Siri | 2011 | Voice assistant launched on the iPhone 4S; uses a mix of on-device speech recognition (since iOS 15) and cloud NLU; rebuilt for Apple Intelligence in 2024 with new architecture and ChatGPT handoff [19][20]. |
| Core ML | 2017 | System-level on-device ML inference framework; supports CNNs, RNNs, transformers, tree ensembles, SVMs, and the .mlpackage format [25]. |
| Vision framework | 2017 | High-level image and video analysis API: face detection, text recognition, object tracking, body pose. |
| Natural Language framework | 2017-2018 | Tokenisation, language identification, named entity recognition, sentiment analysis. |
| Sound Analysis framework | 2019 | Sound event classification (laughter, applause, custom sounds). |
| Create ML | 2018 | Mac-based no-code ML training tool for image, text, sound, activity, and tabular models. |
| Apple Neural Engine | 2017+ | Custom NPU integrated into A-series, M-series, and S-series silicon, growing from 0.6 TOPS on the A11 to 38 TOPS on the M4 [25][31]. |
| Computational photography (Smart HDR, Deep Fusion, Photonic Engine) | 2018-2022 | Per-pixel ML pipelines that run on the Neural Engine to fuse exposures and improve low-light, texture, and skin tones. |
| Live Text and Visual Look Up | 2021 | On-device OCR across the camera and Photos, plus identification of plants, animals, landmarks, and art. |
| Personal Voice | 2023 | Accessibility feature that trains a synthetic copy of the user's voice on-device. |
| Live Captions | 2022 | On-device speech-to-text overlay for any audio. |
| Health and Fitness ML | 2017+ | ECG analysis, fall detection, walking steadiness, sleep stage classification on Apple Watch S-series silicon. |
| Apple Intelligence | 2024+ | The user-facing brand for Apple's generative-AI suite (Writing Tools, Notification Summaries, Image Playground, Genmoji, the new Siri) [33][34][35]. |
| Foundation Models framework | 2025+ | Public Swift API that lets third-party apps call Apple's on-device 3B foundation model with guided generation and tool calling [9][38]. |
| Private Cloud Compute (PCC) | 2024+ | Apple-silicon-based cloud infrastructure for larger Foundation Model requests, with end-to-end attestation [8][41]. |
Apple's AI strategy is inseparable from its silicon strategy. Since the A4 in 2010, Apple has designed its own ARM-based SoCs, all fabricated by TSMC. The integration of CPU, GPU, and the Neural Engine onto a single die, fed by a unified memory architecture, gives ML workloads two structural advantages over discrete-GPU PCs: no PCIe bus to traverse, and the same physical RAM accessible to all three compute units without copies [29][42].
| Family | First chip | Year | First device | Notes |
|---|---|---|---|---|
| A-series (iPhone, older iPad, Apple TV) | A4 | 2010 | iPhone 4, iPad 1 | First Apple-designed SoC. |
| A-series with Neural Engine | A11 Bionic | 2017 | iPhone 8, 8 Plus, iPhone X | First Apple Neural Engine, 0.6 TOPS, 2 cores. |
| Apple Intelligence-class A-series | A17 Pro | 2023 | iPhone 15 Pro, 15 Pro Max | 35 TOPS ANE; minimum supported iPhone for Apple Intelligence. |
| Latest A-series | A18 / A18 Pro | 2024 | iPhone 16 family | Designed around Apple Intelligence; 35 TOPS ANE. |
| M-series (Mac, iPad Pro) | M1 | November 2020 | MacBook Air, MacBook Pro 13, Mac mini, iPad Pro | Began the Mac transition off Intel; minimum Apple Intelligence baseline (with 8+ GB unified memory). |
| Latest M-series | M4 family | 2024 | iPad Pro (May 2024), MacBook Pro (November 2024), iMac, Mac mini | 38 TOPS ANE, marketed as built for Apple Intelligence. |
| Largest M-series | M3 Ultra | 2025 | Mac Studio (March 2025) | Up to 512 GB unified memory; can run 600B-class models locally with quantisation. |
| S-series (Apple Watch) | S1 | 2015 | Apple Watch (1st gen) | Later generations include Neural Engine cores for health features. |
The minimum supported configuration for Apple Intelligence on the iPhone is the iPhone 15 Pro / Pro Max (A17 Pro), and on the Mac and iPad it is any M-series device with at least 8 GB of unified memory [43]. The 8 GB floor is binding because the on-device foundation model and its task adapters need to be paged in alongside the rest of the operating system at all times.
Apple Intelligence is the umbrella brand Apple introduced at WWDC on June 10, 2024 for its generative-AI features in iOS 18, iPadOS 18, macOS Sequoia, and visionOS 2 [33][34]. It first shipped to users on October 28, 2024 in iOS 18.1 in US English [35], with additional English locales in December 2024 and other languages (French, German, Italian, Japanese, Korean, Brazilian Portuguese, Spanish, simplified Chinese) added through 2025.
Apple Intelligence is built on three layers:
The initial Apple Intelligence feature set included Writing Tools, Notification Summaries, smart Mail and Messages reply suggestions, the Photos clean-up tool, Image Playground, Genmoji, and the first stage of the Siri rebuild. The deeper personalised Siri features, with on-screen awareness and cross-app actions, were demoed at WWDC 2024 but delayed in March 2025 to spring 2026 after Apple shifted to a re-architected V2 implementation [36][37]. Apple released the Foundation Models framework at WWDC 2025, exposing the on-device 3B model to third-party app developers through a Swift API with guided generation and tool calling [9][38].
Apple Machine Learning Research operates as a publication outlet for the company's ML teams. Apple papers appear regularly at NeurIPS, ICML, CVPR, ACL, EMNLP, and Interspeech, although the company's volume of published work is smaller than Google DeepMind or Microsoft Research. Notable Apple research outputs include:
apple/ml-ane-transformers repository.Apple's AI strategy can be summarised in four points. First, the company prioritises on-device inference over cloud inference wherever possible; the same engineering team that built Face ID, Live Text, and Visual Look Up also built the on-device foundation model in Apple Intelligence, and Apple's marketing emphasises that user data does not leave the phone in the common case [8][33]. Second, Apple bets on vertical integration: it controls the silicon (Apple Silicon and the Neural Engine), the runtime (Core ML and the Foundation Models framework), and the consumer surfaces (Photos, Camera, Siri, Mail, Notes, Messages). No other large AI player ships at this level of stack control to the consumer.
Third, Apple has been a late mover in generative AI. It did not launch a general-purpose chatbot to compete with ChatGPT, Gemini, or Claude. The Apple Intelligence rollout in 2024 came roughly two years after the public ChatGPT launch in November 2022, and the personalised Siri features were delayed by another year [36][37]. Cook has publicly defended the slower pace, arguing Apple is willing to ship later in exchange for fewer hallucinations and better privacy guarantees. Fourth, Apple is willing to partner where the on-device model is not enough. The OpenAI partnership announced at WWDC 2024 lets Siri hand off a request to ChatGPT, with explicit user permission and IP-address obscuring; ChatGPT integration shipped in iOS 18.2 in December 2024 [33][47]. The 2026 deal with Google routes a custom 1.2-trillion-parameter Gemini model into the next-generation Siri, reportedly paying Google about $1 billion a year [39][40].
| Partner | Year | Nature |
|---|---|---|
| OpenAI | 2024 | ChatGPT integration in iOS 18, iPadOS 18, and macOS Sequoia, with optional GPT-4 access from Siri and the system-wide Writing Tools [33][47]. |
| 2026 | Multi-year licensing of a custom 1.2T-parameter Gemini model to power the rebuilt Siri and other Apple Foundation Models features, run inside Private Cloud Compute, reportedly worth ~$1 billion per year [39][40]. | |
| Anthropic | (negotiated) | Reportedly tested by Apple as a Siri provider before the Google deal; pricing was a sticking point [40]. |
| Microsoft / Perplexity | (rumoured) | Reported by Bloomberg and the Wall Street Journal as alternative providers Apple evaluated. |
| Baidu, Alibaba, Tencent | 2025-2026 | Discussions reported around providing AI features for Apple Intelligence in mainland China, where OpenAI and Google services are restricted. |
| Company | Model strategy | Cloud vs device focus | Headline AI assistant | Headline AI silicon | Approx. market cap (early 2026) | First major generative-AI consumer launch |
|---|---|---|---|---|---|---|
| Apple | Closed; on-device 3B + server PT-MoE; partners with OpenAI and Google | Device-first | Siri (rebuilt 2024) | Apple Silicon + Apple Neural Engine | ~$4T | Apple Intelligence (October 2024) |
| Mixed; closed Gemini family + open Gemma | Cloud-first, with Gemini Nano on-device | Gemini and Google Assistant | TPU + Tensor (Pixel) | ~$2.5T | Bard / Gemini (March 2023) | |
| Microsoft | Heavy partner with OpenAI; also small open Phi family | Cloud-first via Azure, Copilot+ PCs on-device | Microsoft Copilot | Maia + Cobalt + NVIDIA partnerships | ~$3.5T | Bing Chat (February 2023) |
| Meta | Open-weight Llama family | Cloud + on-device (Quest, Ray-Ban Meta) | Meta AI | Custom MTIA + heavy NVIDIA fleet | ~$1.5T | Meta AI / Llama 2 (July 2023) |
| Amazon | Mixed; Nova family + heavy Anthropic stake | Cloud-first via Bedrock | Alexa (Alexa+) | AWS Trainium, Inferentia | ~$2.4T | Alexa LLM, Nova family (2023-2024) |
Market capitalisation figures are approximate and reflect the early-2026 reporting environment around the Cook succession announcement [5][15].
Apple's AI program has drawn three recurring lines of criticism.
The first is delivery risk. The personalised Siri features demoed at WWDC 2024 and used in iPhone 16 marketing were postponed in March 2025 to spring 2026, with Apple acknowledging that the V1 architecture could not meet quality bars [36][37]. Customers who upgraded specifically for the demoed features were unhappy, and Apple was hit with class-action lawsuits over the marketing in early 2025.
The second is the gap between Siri and current chatbots. Independent reviewers have consistently rated Siri behind ChatGPT, Gemini, and Claude on reasoning, summarisation, and follow-up benchmarks, even after the Apple Intelligence overhaul. The 2026 Google partnership is partly an admission that Apple's own foundation models are not yet sufficient for the conversational Siri Apple wants to ship.
The third is structural. Apple is heavily reliant on TSMC for silicon manufacturing, on Foxconn and other Chinese suppliers for assembly, and on a small set of US, Indian, and Vietnamese facilities for diversification. Tariff and geopolitical risk through 2025 and 2026 has weighed on the stock and complicated the Vision Pro launch, widely characterised in the press as a slow start. Apple's privacy-first design also creates an ongoing tradeoff: features that require very large models are awkward to ship on a 3B on-device model and a constrained PCC environment. The Google and OpenAI partnerships are the company's pragmatic answer.
The Apple Intelligence rollout that began in October 2024 continued through 2025 with new languages, additional features such as visual intelligence (a camera-based search and information overlay launched on iPhone 16 in late 2024), and the Foundation Models framework for third-party developers at WWDC 2025. The personalised Siri delay was confirmed in March 2025, with a target of iOS 26.4 in spring 2026 [36][37]. In November 2025, Bloomberg reported that Apple was finalising a deal to license a custom 1.2-trillion-parameter Gemini model from Google to back the rebuilt Siri; the partnership was confirmed jointly on January 12, 2026 [39][40]. In April 2026, Apple announced that Tim Cook would step down on September 1, 2026, with John Ternus taking over as CEO [6]. Apple Intelligence and the underlying foundation models remain in active development, and the company's bet on on-device inference, vertical integration, and selective partnership continues to define its position in the AI industry.