This article organizes the annual AI trends report from Bond Capital, founded by internet queen Mary Meeker, which vividly illustrates the panoramic view of how AI technology is rapidly reshaping the world. (To be organized into three articles)
(Background: Foxconn’s internal experiment: AI can replace 80% of jobs, Chairman Liu Yangwei reveals a trinity blueprint for future factories)
(Background Supplement: Podcast Highlights: AI and robots cover the cryptocurrency craze, the next era of micro-entrepreneurship is the hottest)
Known as the “Queen of the Internet,” Mary Meeker’s Bond Capital is globally recognized for its annual Internet Trends report. Recently, BOND released its latest report titled “Trends – Artificial Intelligence,” which contains 340 pages that profoundly depict how AI technology is reshaping the world at an astonishing pace. From the disruptive changes in user habits, unprecedented capital expenditures, to fierce competition in business models and the far-reaching effects on the physical world and future jobs, it reveals the astonishing acceleration of AI development. Due to the extensive content of the report, the Blockchain district will summarize the key points related to AI in three articles, of which this is the first; please click here for the second.
A New “Year of the Dog” in AI Technology,
Even Faster
The Bond report opens with a compelling tone, referencing Vint Cerf’s observation in 1999 that “a year in the Internet industry is like a year in dog years.” BOND further points out that the evolution of artificial intelligence is “accelerating at an even faster rate… and the capabilities of machines are surpassing us.” BOND states that this unprecedented pace of development is not mere rhetoric, but is supported by a substantial and continuous upward trend in user numbers, usage, and revenue charts, as well as similarly rising expenditure charts. This is not just a quantitative accumulation but a qualitative leap, indicating that a new technological paradigm is forming.
The global proliferation of internet infrastructure (covering 5.5 billion citizens), the vast digital datasets accumulated over decades, and the emergence of groundbreaking large language models (LLMs)—especially OpenAI’s ChatGPT, which has rapidly gained popularity since its launch in November 2022 due to its “extremely user-friendly and responsive interface.” The BOND report emphasizes that the world is undergoing unprecedented transformation, with artificial intelligence being the core engine of this change. ChatGPT is hailed as “the most successful overnight sensation in history (nine years after its establishment),” its impact being a stark contrast to the gradual spread of the early Internet, which almost swept the globe instantaneously.
This phenomenon has ignited fierce competition between existing market leaders and emerging challengers, all racing to build the next generation of AI infrastructure, including more autonomous agent interfaces and sovereign AI models. The intertwining of technological advancements and geopolitical forces is an important theme that runs throughout the report.
Is change faster than ever before? The answer is yes.
The Compounding Effect of Technology
Millennial Dimension: A retrospective on over a thousand years of global GDP growth clearly shows a close correlation between significant technological advancements (such as printing, steam engines, electrification, the Internet, and cloud computing) and notable economic expansion. Each technological wave has laid the groundwork for subsequent developments, exhibiting an accelerating trend.
Continually Expanding Scale: The development cycle of computers is continuously shortening, while the scale continues to expand. From the early million-unit mainframes to the potential market of “hundreds of billions of devices” targeted in the current AI era, this exponential growth is clearly visible. The technological foundation of this process has evolved from the initial CPUs to the combination of big data and cloud computing, and now to the current era centered around GPUs.
Multidimensional Explosive Growth of Artificial Intelligence:
- Training Data Scale: In the past fifteen years, the amount of text data used to train key AI models has grown at an astonishing annual rate of 260%. This means that models can learn from a broader and deeper pool of knowledge, enhancing their understanding and generative capabilities.
- Training Computational Power (FLOPs): Growing alongside data scale is the computational power required for training. Over the past fifteen years, the FLOPs used for training key AI models have achieved an annual growth rate of 360%, making the emergence of larger and more complex models possible.
- Algorithm Improvement: In addition to the increases in data and computational power, advancements in the algorithms themselves are also crucial. In the past nine years, smarter algorithms have enabled effective computational power to grow at an annual rate of 200%, meaning that the same or even better results can be achieved using fewer resources.
- Performance of AI Supercomputers: The performance of dedicated AI supercomputers has also rapidly advanced, achieving an annual growth rate of 150% over the past six years (measured in FLOP/s).
- Number of Powerful AI Models: The number of newly released large-scale AI models (defined as models requiring more than 10^23 FLOPs of training power) has seen an annual growth rate of 167% over the past four years.
ChatGPT: A Reflection and Catalyst of Acceleration:
The report regards the rise of ChatGPT as a landmark event in this current wave of AI acceleration, with its rapid increases in user numbers (reaching 800 million weekly active users by April 2025), subscribers, and revenue being particularly noteworthy.
ChatGPT achieved an average of 365 billion searches per year in just two years (by 2024), whereas it took Google 11 years (until 2009) to reach this milestone, making it 5.5 times faster. This stark contrast vividly highlights the rapid iteration and market penetration of AI technology.
This clearly illustrates how artificial intelligence serves as a “compounding engine on internet infrastructure,” achieving explosive growth based on existing technological foundations.
Profound Evolution of Knowledge Dissemination:
The report divides the history of knowledge dissemination into three stages:
- From 1440 to 1992, the “static + physical dissemination” stage represented by printing.
- From 1993 to 2021, the “dynamic + digital dissemination” stage characterized by the Internet.
- From 2022 to the present, the new era of “dynamic + digital + generative dissemination” represented by generative AI such as ChatGPT.
This evolutionary process is not just a technological innovation but a fundamental transformation in the way information is acquired, processed, and created. The report quotes Martin H. Fischer’s famous saying: “Knowledge is the process of piling up facts; wisdom lies in simplifying them,” which is especially apt in the era of generative AI, as AI simplifies complex information, giving new vitality to knowledge.
The Development History of Artificial Intelligence: Key Milestones and Expanding Capability Boundaries:
The report traces a detailed “timeline of AI milestones,” from Alan Turing’s proposal of the Turing Test in 1950 to predictions for 2025. It includes events such as the “AI winter” (1967-1996), the victory of Deep Blue, the popularization of Roomba vacuum robots, the birth of Siri voice assistants, various iterations of GPT models, and significant releases from tech giants like Anthropic, Meta, Google, and Alibaba.
Unprecedented Trends in Artificial Intelligence Development:
- Industry Surpassing Academia: Around 2015, with the sharp increase in demand for data, computational power, and funding, the industry surpassed academia in releasing influential machine learning models, becoming the main driver of AI innovation.
- Flourishing Developer Ecosystem: Taking the NVIDIA ecosystem as an example, the number of developers grew sixfold within seven years, reaching 6 million. Similarly, Google’s Gemini ecosystem reported a fivefold increase in developers, reaching 7 million. This vast developer community is the cornerstone of AI applications and innovations.
- Surge in Patent Applications: The number of computing-related patent grants in the U.S. has experienced two significant explosive growth phases: the first after Netscape went public in 1995, and the second after ChatGPT was released in 2022, with the latter showing even more rapid growth.
- Continuous Breakthroughs in Performance: In 2024, AI systems first exceeded the average human level in the MMLU (Massive Multi-task Language Understanding) benchmark test. In a Turing test experiment in the first quarter of 2025, as many as 73% of AI responses were mistakenly identified by testers as human-generated, demonstrating significant advancements in AI’s ability to simulate human conversation.
The realism of image generation technology has significantly improved (e.g., the evolution of Midjourney from v1 to v7 and comparisons between AI-generated faces and real faces). Music translation and generation technologies have also matured and become increasingly realistic, with services like ElevenLabs rapidly gaining market acceptance while mainstream platforms like Spotify start integrating AI translation features. A visual overview provided by Morgan Stanley clearly illustrates the diverse emerging applications of AI across various fields.
Opportunities and Risks Coexist Duality
The report also soberly recognizes the duality of AI development and cites the classic work “Artificial Intelligence: A Modern Approach” by Stuart Russell and Peter Norvig. The benefits of AI are evident, as it is expected to amplify human wisdom and address many significant challenges.
However, its potential risks cannot be ignored, including misuse in areas such as autonomous weapons, mass surveillance, biased decision-making, and cybersecurity. Stephen Hawking’s thought-provoking warning still resonates: “Successfully creating AI could be the most significant event in the history of our civilization. But it could also be the last—unless we learn how to avoid the risks.” This statement profoundly reveals the significant choices and responsibilities facing the current development of artificial intelligence.
The first article aims to outline the astonishing speed and multidimensional features of current AI development through the lens of the BOND report. The compounding effects between data, computational power, algorithms, and investment are jointly spawning a technological wave that is unparalleled in speed and global influence.
The next articles will delve deeper into user growth and capital expenditure, cost dynamics, commercialization pathways, competitive landscapes, and the profound impact of AI on the physical world and future work.
》》Complete Report on Trends – Artificial Intelligence
》》340-page in-depth AI report (Part 2): Training expenditures continue to rise + usage costs plummet, the contradictions of AI model economics
