About the paper
The report is a mixed-methods, chart-led secondary analysis of artificial intelligence trends, compiled by BOND from a wide range of public and private company data, market research, government sources and additional non-public insights.
It does not present one unified survey or a single fieldwork sample, so the total number of respondents, cases or participants is not clearly specified in the report; those figures vary by source and chart.
The geographic scope is global, with a strong emphasis on the United States and China, alongside regional and country-level comparisons.
BOND Capital and Mary Meeker have not released AI trend reports in years other than 2025. The 2025 “Trends – Artificial Intelligence” edition marked a revival of Meeker’s trend series after a six-year gap, evolving from her famous annual “Internet Trends” reports that ran from 1995 to 2019
Length: 340 pages
More information / download:
https://www.bondcap.com/reports/tai
Core Insights
1) Why does the report argue that AI change is happening faster than ever?
The report’s central argument is that AI is not simply another technology cycle but a compounding force built on top of existing internet infrastructure, massive digital datasets, better chips, improving models and intense capital deployment. BOND presents AI as accelerating faster than earlier technology waves because the rails were already in place: billions of connected users, decades of data accumulation and a ready-made global digital distribution system. That is why the report repeatedly frames the current moment as unprecedented in pace and scope, spanning technical, financial, social, physical and geopolitical change all at once.
It also argues that AI differs from earlier waves because it arrived into a world that was already organised, connected and digitally mature. In the report’s telling, the internet took years to build the conditions for mass diffusion, whereas generative AI could scale immediately on top of them. This makes AI both a product of prior technological compounding and a new multiplier on top of it. The report therefore treats AI as a step-change in how information is accessed, created and distributed, rather than as a simple continuation of software history.
A further reason the report sees change as unusually fast is that the race is not only commercial but geopolitical. It repeatedly links AI progress to strategic competition, especially between the US and China, and suggests that this rivalry is intensifying investment, product release cycles and the urgency of deployment. In that sense, speed is presented not just as a market phenomenon but as a consequence of state-level and corporate competition feeding each other.
2) What evidence does the report provide that AI adoption and usage are scaling at an exceptional rate?
The report uses ChatGPT as its clearest proxy for AI adoption and argues that its scale-up has been historically extraordinary. One headline figure is that ChatGPT reached an estimated 800 million weekly active users by April 2025. Another is that BOND compares ChatGPT’s path to 100 million users with earlier consumer platforms and concludes that it reached that threshold far faster than services such as Netflix, LinkedIn, Instagram and TikTok. The argument is straightforward: AI is not merely growing quickly; it is outpacing the adoption curves of landmark digital products.
The report also stresses how global that adoption is. One of its most striking comparisons is that around 90% of ChatGPT app users were outside North America by year three, whereas the internet took roughly 23 years to reach a similar share. That comparison underpins one of the report’s biggest claims: unlike earlier foundational technologies that spread outwards from the US over a long period, AI reached a global audience almost immediately. The geographic focus here is explicitly worldwide, though the report notes that availability constraints in places such as China affect the underlying app data.
Beyond end users, the report points to rising developer adoption and ecosystem participation. It highlights 6 million developers in NVIDIA’s ecosystem and says Google reported more than 7 million developers building with Gemini, up fivefold year on year. This matters because the report is not only describing consumer enthusiasm; it is also arguing that the builder base around AI is widening quickly, which in turn supports more products, more infrastructure demand and more downstream usage.
3) How does the report explain the economics of AI, especially the tension between rising investment and falling usage costs?
A key theme in the report is that AI economics are pulling in two directions at once. On one side, training and infrastructure costs are high and still rising. BOND highlights very large capital expenditure by major US technology firms and shows the “Big Six” reaching $212 billion in capex in 2024, up 63% year on year. That supports the report’s broader claim that AI requires enormous spending on compute, data centres and model development.
On the other side, the report argues that the cost of using AI is falling sharply for customers and developers. Its framing is that while frontier training remains expensive, inference is becoming cheaper and cheaper, which leads to wider access, more experimentation and increasing convergence in model performance. In other words, the barriers to building frontier systems remain high, but the barriers to using AI tools are falling. That combination is central to the report’s optimism about continued adoption.
The report also makes clear that monetisation remains unresolved. One illustrative chart compares estimated revenue and compute expense for a leading US-based AI LLM company and shows strong revenue growth alongside very large compute losses. That supports a more nuanced economic reading: AI demand is real, but profitability is not yet settled. The report therefore presents the current phase as one in which consumers and enterprise users are benefiting from rapid improvement and falling usage costs, while producers are still absorbing heavy investment burdens.
4) What does the report suggest about competition, monetisation and the global balance of power in AI?
The report presents competition in AI as both fierce and structurally destabilising. It emphasises that incumbents, startups, open-source communities and state-backed ecosystems are all competing at once. One reason monetisation looks fragile, in BOND’s view, is that open-source momentum and Chinese advances are placing pressure on the pricing power and defensibility of leading US model providers. The report therefore does not portray the current leaders as secure winners. Instead, it describes a market in which advantage can erode quickly.
That point becomes especially clear in its competitive charts. BOND shows relative desktop user-share shifts across leading LLMs and highlights the emergence of a Chinese model within a short period. It also repeatedly returns to US–China rivalry as a defining strategic frame, suggesting that AI leadership may translate into broader geopolitical influence. In this sense, the report sees AI not just as a commercial contest over products and margins, but as a contest over standards, platforms and long-term international dependence.
At the same time, the report is not purely alarmist. Its perspective is broadly pro-innovation and cautiously optimistic. It argues that intense competition may accelerate progress, widen access and keep the field dynamic, even if it also increases uncertainty. But its assumptions are clear: AI is now a strategic domain, leadership matters, and no company or country can assume its position is fixed.
5) What broader implications does the report identify for work, society and the physical world?
The report argues that AI is moving beyond chatbots and software assistance into the “physical world” and the workplace. One example it uses is autonomous mobility in San Francisco, where an autonomous taxi provider’s operating-zone market share rises sharply over the period shown. That chart is meant to illustrate that AI is no longer confined to digital interfaces; it is beginning to reshape real-world services, logistics and robotics-related applications.
On work, the report says change is already visible rather than merely speculative. A chart on US IT jobs shows AI-related postings rising strongly while non-AI postings decline over the same indexed period. The implication is not simply that more technology jobs are appearing, but that the composition of demand is changing. BOND’s broader argument is that AI is likely to reconfigure the nature of work, the skills that employers value and the kinds of roles that expand or contract.
The report also addresses social implications in a wider sense. It includes a benefits-and-risks section that acknowledges the promise of AI for productivity, science and material abundance, while also flagging risks such as surveillance, persuasion, employment disruption, biased decision-making, cybersecurity problems and safety-critical misuse. That matters because the report is not making a narrow market-growth case. It is arguing that AI is becoming a civilisation-scale force whose effects will be economic, political and societal at the same time.
Taken together, the report’s conclusion is that AI’s significance lies in its breadth. It is changing information flows, business competition, labour demand, public power and physical systems all at once. That is why BOND treats this moment less as a discrete technology story and more as a large-scale transformation in how economies and societies operate.