In November 2025, noted technology analyst Benedict Evans published his annual macro‑trend presentation "AI Eats the World". This nearly 200‑page deck continues Evans' signature approach, using charts, historical analogies and strategic insight to examine how artificial intelligence (AI) is reshaping the global economy, technology platforms and everyday life. Evans, a former Andreessen Horowitz partner and independent analyst, draws millions of readers each year and his reports are regarded as an industry bellwether. Unlike his earlier focus on "Mobile Eats the World," this report emphasizes AI as the next platform‑level force, but Evans is cautiously optimistic: AI is not an instant magic, but will permeate gradually like electricity or the internet.
Authoritative analyses on Substack and Medium note that the report's core is to reveal the gap between AI hype and reality while offering a practical framework to help companies and individuals navigate this transition. This special report is based on Evans' original deck and related summaries, distilling its main points into sections on the AI paradox, productivity transformation, platform displacement, risks and forecasts, with the aim of providing readers factual and actionable insight.
AI paradox: Everyone builds it, few use it
Evans opens the report by pointing to a major "AI paradox": despite intense investment and development, actual adoption lags. In 2025, AI startup funding exceeded $200 billion and firms such as OpenAI saw valuations above one trillion, yet enterprise applications remain largely experimental. Evans supports this with data: only 15% of global companies have achieved scaled AI deployment, far below mobile app adoption in 2020. He draws a historical parallel: after the invention of electricity, factories took 30 years to redesign production lines; similarly, although AI is powerful, companies must redesign workflows rather than simply "plug in" models.
Evans attributes the paradox to "expectation inflation": media portrayals cast AI like ChatGPT as a "cure‑all," but in reality it shines in creative tasks (such as generating code) while facing challenges in structured work because of data privacy and accuracy issues. Evans predicts this paradox will gradually resolve by 2030, as AI becomes integrated into Office and CRM tools and enterprise adoption doubles. Substack summaries note that this view reminds practitioners: AI is a starting point, not an endpoint, and building the ecosystem takes time.
Productivity transformation: Reshaping work, not wholesale replacement
One of the report's core arguments is AI's impact on productivity: it will reshape workflows and boost efficiency, but will not broadly replace humans. Evans cites McKinsey data showing AI could automate 45% of office tasks by 2030 while creating more high‑value roles such as AI ethicists and data curators. He compares it to the automobile: early on it displaced coachmen but created auto engineers and road designers. Likewise, AI amplifies human capability in coding, design and analysis: for example, GitHub Copilot has helped developers save about 30% of their time.
Evans emphasizes that productivity gains depend on "last‑mile" integration: current models are good at generation but require human verification and iteration. In the deck he shows that AI can increase output 2–3x in creative industries (such as ad generation) but achieves only about 85% accuracy in medical diagnosis, necessitating expert oversight. Medium analyses argue this challenges the "AI unemployment" narrative and instead highlights skills upgrading: by 2027, roughly 100 million people worldwide may need training to use AI tools. Evans predicts AI will become as ubiquitous in the office as Excel and could boost GDP by 1–2%.
Platform displacement: Who controls AI distribution?
Evans frames AI as the "next platform," akin to iOS or Windows, but the battle over distribution is fierce. The report analyzes OpenAI's ChatGPT model (a closed ecosystem) versus Google's Gemini (search integration)—the former controls the user interface while the latter controls data flows. He likens it to the smartphone wars: Apple dominated distribution through the App Store, and the AI platform battle will center on an "AI OS," with contenders like Microsoft's Copilot or upgraded versions of Apple's Siri.
Key data: in 2025 AI API call volume reached the trillions, but 90% of traffic was controlled by a handful of platforms. Evans warns that while open‑source models (such as Llama) have emerged, commercialization still faces distribution bottlenecks. Reddit discussions framed this as an "AI cold war": competition among giants will determine who controls a trillion‑dollar market. Evans predicts that by 2035 AI platforms will reshape advertising, search and e‑commerce in a way comparable to Google's PageRank revolution.
Risks and forecasts: Opportunities with caveats
The report does not shy away from AI risks: bias, privacy and employment impacts. Evans cites Stanford data indicating bias rates against women and minority groups in AI models reach 20%, requiring more diverse training data to mitigate. He compares AI to nuclear power: its potential is enormous, but lagging regulation could cause crises such as widespread deepfakes or algorithmic discrimination. On employment, the report foresees short‑term automation of low‑skill jobs but long‑term creation of new opportunities, similar to the industrial revolution.
In its forecasts Evans is optimistic: AI will "eat the world," but progressively—2025–2030 will be an integration phase, with acceleration after 2030. He uses the "electricity factory" analogy: early efficiency gains are limited, then a later restructuring changes everything. A Metaschool summary's five key takeaways reinforce this: heavy AI investment needs time for ROI; the platform war will decide winners; and risks require governance.
Industry impact and takeaways
Evans' report has had significant influence: YouTube summary videos have millions of views and countless LinkedIn shares. It offers a framework for companies: invest in AI integration rather than chasing hype. HT‑X called it an "operations manual" for navigating the AI world.
In short, "AI Eats the World" summarizes the AI paradox, productivity shifts, platform displacement and risks, stressing a gradual transformation. Evans' data‑driven perspective helps readers avoid hype and seize opportunities. In the future, AI is likely to reshape the world as Evans predicts, but cautious, methodical progress is required.
▶️ Enjoyed this piece? Creating content is hard!
Follow our YouTube channel:
AI Rosetta
Your likes, subscriptions and shares are our greatest support! 🙏
More AI news at: https://www.airosetta.com