On a Saturday afternoon, I decided to walk from Vila Conceição to Ibirapuera Park with Dogfight in hand. At first, the book felt like a return to a closed chapter in technology history. A few pages later, it started to read less like a historical record and more like a structural map of the current AI war.
That is what still makes Fred Vogelstein’s Dogfight useful. The book matters not simply because it narrates the clash between iPhone and Android. It matters because it shows what happens when a new interface becomes the main gateway to computing. Even the publisher’s description frames the story in broader terms: not only which devices would replace phones and laptops, but who would control content and distribution on the next generation of devices. That same logic has returned at a larger scale in the 2020s. The smartphone was the strategic object of the previous cycle. The current cycle revolves around models, chips, cloud, data centers, energy, operating systems, distribution, and interface.
Why Dogfight Still Matters in the AI War
At first glance, Dogfight reads like a story of corporate rivalry. A closer reading reveals something more structural: platform conflict. The smartphone was the visible object, but the real battle centered on interface control and the access layer between users, software, content, and markets. That is why the book still holds up. Its enduring value is not the 2013 product lineup. It is the grammar of competition: platforms beat isolated products, interfaces redefine winners, distribution matters as much as engineering, and partners become rivals when the dominant layer changes.
The final stretch of the book matters as much as the opening. The lens widens beyond devices and operating systems into media, content, and distribution. That widening is important because it reveals what the mobile battle was already becoming: a fight over access, attention, and control of channels. If Mark Zuckerberg appears in that historical frame, the correct name is Facebook, not Meta. Dogfight was published in 2013, long before Meta’s corporate rebrand. The connection to Meta AI and Llama 4 belongs to the present analysis, not to the original text. Meta introduced Llama 4 in April 2025 and launched the Meta AI app later that month.
A Compressed Timeline of the AI War
A short timeline helps show how quickly the AI war expanded. ChatGPT launched on November 30, 2022, turning the conversational interface into a mass-market product. Claude followed on March 14, 2023. Microsoft introduced Microsoft 365 Copilot on March 16, 2023, moving AI directly into enterprise workflow. Amazon announced Bedrock on April 13, 2023, formalizing AWS’s position in model infrastructure and distribution. xAI announced Grok on November 3, 2023. Google introduced Gemini on December 6, 2023 as a multimodal model family. OpenAI revealed Sora on February 15, 2024 as a major signal in generative video. Apple introduced Apple Intelligence on June 10, 2024 as a personal intelligence system integrated across its devices. Meta introduced Llama 4 on April 5, 2025 and launched the Meta AI app on April 29, 2025.
This sequence shows how quickly the conflict expanded. In less than three years, the center of gravity moved from a single chatbot launch to a distributed contest spanning interface, workflow, cloud, social platforms, multimodality, and infrastructure.
From Device War to AI Infrastructure Race
The earlier battle was fought over the mobile interface. Today’s conflict is centered on the cognitive interface. In the mobile era, the core stack was touchscreen, operating system, and app store. In the AI era, the core stack includes chat, copilots, agents, voice, persistent context, model-mediated search, and assisted execution. The home screen organized apps. The new layer organizes context, inference, and execution. This is why Dogfight still matters in the current AI war. The names and products changed, but the structural question did not: who controls the main layer through which users access software, information, and action?
The AI war is no longer centered on the device alone. It is a cognitive and physical stack made of models, chips, cloud, data centers, energy, operating systems, interface, and distribution. NVIDIA’s language is revealing here: it describes “AI factories” as integrated systems designed to produce intelligence at scale. That formulation captures the shift. Intelligence is no longer only a software feature. It is an infrastructure product.
That is also why xAI matters in this article. Grok is the visible interface, but the larger point is that xAI pairs product with compute infrastructure. Its current positioning only makes sense when model, interface, and supercompute are treated as one strategic unit. At its core, the AI war is also an AI infrastructure race.
If the mobile war depended on hardware and operating systems, the AI war depends on a much heavier stack: models, chips, cloud, data centers, energy, and interface. That is the real scale change between the 2010s and the 2020s.
Where the AI Platform War Is Being Fought
Apple occupies the layer of vertical integration: device, operating system, and personal context. Google occupies distribution at scale across search, Android, productivity, and services. Microsoft’s weight comes from enterprise workflow and the integration of AI into work software. Amazon anchors cloud and enterprise infrastructure through AWS and Bedrock. Meta combines massive social distribution with an open-weight multimodal strategy through Llama and Meta AI. This is where the AI platform war becomes visible.
Alongside the big tech layer sit newer power centers. OpenAI holds the strongest mass-market conversational interface. Anthropic is especially important in agents, vision, and enterprise execution. xAI links model, interface, and compute. NVIDIA anchors the physical infrastructure layer.
Multimodality Inside the AI War
If multimodality appears in the article, it should be handled by function. At the top are stack-level platforms such as OpenAI, Google, Meta, and Anthropic, which combine models with broader ecosystem control. At the applied edge are creative tools such as Runway, Midjourney, Pika, and Sora. These matter because they show that the AI war has already expanded into image, voice, and video. But the centralization point sits one layer above the tool itself. Real concentration happens where model, compute, infrastructure, distribution, and interface converge.
Sora is useful here as a cautionary example. OpenAI has stated that the current Sora web and app experiences will be discontinued on April 26, 2026, and the API on September 24, 2026. The product can change quickly. The stack logic remains.
What Dogfight Still Explains About the AI War
As a factual narrative, Dogfight belongs to its time: smartphones, iOS, Android, and the specific competitive configuration of 2013. As a structural map, it remains highly useful. Platforms still beat isolated products. Interfaces still define power. Distribution still matters as much as visible innovation. Infrastructure still determines who can scale. Alliances remain temporary.
That is the key distinction. The book is time-bound as history, but not as strategy.
Conclusion
Dogfight remains relevant not as nostalgia, but as pattern recognition. It shows what happens when a new interface reorganizes markets, distribution, and power. That happened with the smartphone. It is happening again, at a higher level, with artificial intelligence.
The center of the current AI war is not the model alone. It is the convergence of model, chip, data center, energy, operating system, interface, and distribution. Whoever controls that layer will not simply build a successful product. They will shape how users access systems, how work gets executed, and how digital power is centralized.
Dogfight explained the war for the mobile interface. This decade is defining the war for the cognitive and physical interface.
