Mary Meeker's 2025 AI Report: What the Data Reveals About AI's Unprecedented Growth

Mary Meeker, the legendary technology analyst known as the "Queen of the Internet", released her first comprehensive trends report since 2019 in May 2025. The 340-page analysis focuses entirely on artificial intelligence and reveals growth patterns unlike anything seen in technology history.

Why This Report Matters Now

The pace and scope of change related to artificial intelligence technology evolution is indeed unprecedented, according to Meeker's extensive data analysis. The report uses the word "unprecedented" 51 times across its pages, highlighting how AI adoption is fundamentally different from previous technology cycles.

What makes AI adoption different? Unlike the gradual rollout of internet technology that started in the USA and slowly spread globally, ChatGPT hit the world stage simultaneously, growing in most global regions at once. This represents a completely new pattern of technology diffusion.

The Speed of AI Adoption is Breaking Records

ChatGPT's Historic Growth Trajectory

ChatGPT achieved something no technology product has accomplished before: reaching 800 million weekly active users in just 17 months. To put this in perspective, Netflix took over 10 years to reach 100 million users, while ChatGPT did it in 2 months.

The search volume comparison is equally striking. ChatGPT hit 365 billion annual searches in 2 years, while Google took 11 years to reach the same milestone. This represents a 5.5x faster path to massive scale.

How global is ChatGPT's reach? By year 3, 90% of ChatGPT users were located outside North America. India leads with 14% of global users, followed by the USA at 9% and Indonesia at 6%. This global distribution happened faster than any previous technology platform.

Enterprise and Developer Adoption Accelerating

The developer ecosystem is experiencing explosive growth. NVIDIA's developer ecosystem expanded from minimal participation in 2005 to 6 million developers by 2025, representing 6x growth over seven years. Similarly, Google reports 7 million developers now building with Gemini, a 5x year-over-year increase.

Enterprise adoption is following suit. GitHub Copilot has been adopted by over 77,000 organisations, representing 180% year over year growth. These are not just experimental implementations: companies are seeing real productivity gains and expanding their usage.

The Investment Surge Behind AI Growth

Record-Breaking Capital Expenditure

The "Big Six" technology companies (Apple, NVIDIA, Microsoft, Alphabet, Amazon, and Meta) spent a combined $212 billion on capital expenditure in 2024, representing a 63% year over year increase. This level of spending is unprecedented in technology history.

What's driving this investment? The primary driver is AI infrastructure buildout. Companies are racing to build data centers, acquire specialized chips, and develop the computing capacity needed for AI training and inference at scale.

Capital expenditure now represents 15% of revenue for these companies, compared to 8% ten years ago. This shift indicates a fundamental reallocation of resources toward AI capabilities.

Revenue Growth Across the AI Stack

The investment is generating returns across multiple layers of the technology stack:

Chip Manufacturers: NVIDIA's data centre revenue grew 78% year over year to $39 billion. The company now captures 25% of global data centre CapEx, up from minimal share just years ago.

Cloud Providers: Microsoft's AI business surpassed $13 billion in annual run rate, representing 175% year over year growth.

AI Model Companies: OpenAI generated $3.7 billion in revenue in 2024, representing a 1,050% annual growth rate. The company reached 20 million paid subscribers.

How AI Performance is Changing Market Dynamics

The Great AI Model Convergence

One of the most significant trends identified in Meeker's report is the convergence of AI model performance. The gap between the highest performing models and alternatives is rapidly shrinking.

What does performance convergence mean? DeepSeek R1, an open source model from China, scored 93% on math benchmarks compared to OpenAI's o3-mini at 95%. For many practical applications, this difference is negligible.

Simultaneously, inference costs have plummeted 99.7% over two years. This combination of improving performance and falling costs is democratising access to advanced AI capabilities.

The Rise of Open Source Competition

China is leading the open source AI movement, releasing three large scale models in 2025 compared to US competition. Chinese models like DeepSeek have grown from 0% to 21% global market share in just months.

Why does open source matter? Open source models can be downloaded, modified, and deployed without ongoing API costs. Meta's Llama models have been downloaded over 1.2 billion times, with most downloads representing derivative versions created by developers.

AI's Physical World Impact

Autonomous Systems Scaling Rapidly

AI is moving beyond digital applications into physical systems with measurable impact:

Autonomous Vehicles: Tesla has accumulated over 4,000 million cumulative self-driving miles, representing a 100x increase over 33 months. Waymo now captures 27% of San Francisco rideshare bookings, growing from 0% in just 20 months.

Industrial Applications: Carbon Robotics has used AI-powered laser systems to weed over 230,000 acres, preventing the use of 100,000+ gallons of herbicides. Applied Intuition now serves 18 of the world's top auto manufacturers with AI-powered vehicle development tools.

Infrastructure Buildout at Unprecedented Speed

The report highlights xAI's Colossus facility as an example of accelerated AI infrastructure deployment. The facility went from 0 to 200,000 GPUs in just 7 months, built in a 750,000 square foot facility completed in 122 days—half the time it takes to build an average American house.

Global Competition and Market Leadership

The US-China AI Race

The report identifies AI leadership as potentially decisive for geopolitical influence. As Meta CTO Andrew Bosworth noted, "This is our space race... there's very few secrets. And you want to make sure that you're never behind."

How are the two countries positioned? The USA leads in proprietary model development and has 70% of the top 30 technology companies by market capitalisation. China dominates in open source model releases and has more industrial robots installed than the rest of the world combined.

Chinese AI users show markedly different sentiment toward AI adoption. 83% of Chinese respondents believe AI products have more benefits than drawbacks, compared to only 39% of US respondents.

Market Share and User Preferences

Global usage patterns reveal interesting regional preferences:

Closed vs. Open Models: Closed source models like ChatGPT and Gemini dominate consumer usage globally, while open source models gain traction among developers and in specific regions.

Regional Champions: China's top 10 AI apps by monthly active users are all domestically developed, while global platforms dominate elsewhere.

Workforce and Economic Implications

Changing Nature of Work

AI adoption is reshaping how work gets done across industries. The report shows AI job postings have increased 448% over seven years, while non-AI IT job postings declined 9%.

What types of jobs are emerging? Companies are creating entirely new categories of AI-related positions. Apple alone has over 600 job openings related to generative AI as of May 2025.

Enterprise adoption surveys show 7% of US firms are using AI as of Q1 2025, with 21% quarterly growth in adoption. The focus is primarily on productivity improvements rather than cost reduction.

Productivity Impact Examples

Real-world productivity gains are being documented across industries:

Healthcare: Kaiser Permanente has completed over 10 million AI-assisted patient visit summaries, with doctors reporting significant time savings.

Financial Services: JP Morgan estimates 35-65% value increases from AI/ML implementations across various business functions.

Customer Service: Bank of America's Erica virtual assistant has handled 2.5 billion customer interactions since its 2018 launch.

What the Data Reveals About AI's Future

Infrastructure Requirements Continue Growing

Data centres now account for 1.5% of global electricity consumption, with growth rates of 12% annually since 2017 (four times faster than total electricity consumption growth). The USA accounts for 45% of global data centre electricity usage.

Energy implications: The report notes that AI's power demands are increasing, but AI is also enabling energy efficiency gains through optimisation and management systems.

Investment vs. Revenue Realities

The report presents a nuanced view of AI economics. While revenue growth is impressive, many AI companies are burning significant capital. OpenAI's compute expenses reached $5 billion against $3.7 billion in revenue for 2024.

The venture capital perspective: Private AI companies have raised approximately $95 billion against roughly $11 billion in annualized revenue, representing high valuation multiples that reflect both opportunity and risk.

Key Questions for Business Leaders

Q: How fast should companies adopt AI capabilities? A: The report suggests urgency is critical. As Shopify CEO Tobias Lütke noted, "Reflexive AI usage is now a baseline expectation." Companies that move slowly risk falling behind competitors who integrate AI capabilities faster.

Q: Should businesses focus on proprietary or open source AI models? A: The choice depends on use case and resources. Open source models offer cost advantages and customisation options, while proprietary models provide easier implementation and potentially better performance for general applications.

Q: What sectors will see the most AI disruption? A: The data shows AI impact across all sectors, but software development, customer service, healthcare, and financial services are experiencing the most rapid transformation.

Q: How important is AI infrastructure investment? A: Critical for competitive positioning. The report shows companies investing 15% of revenue in AI-related CapEx are gaining market leadership positions.

Q: What is the biggest risk facing AI companies today? A: The economics of high training costs versus falling inference prices. Many AI companies are spending billions on model training while inference costs have dropped 99.7%, creating a challenging path to profitability that only the most efficient operators may survive.

Q: How is AI changing the global technology landscape between countries? A: China and the USA are emerging as the two dominant AI superpowers, with fundamentally different approaches. The USA leads in proprietary models and has 70% of top tech companies, while China dominates open source releases and shows 83% citizen approval for AI benefits versus only 39% in the USA.

Q: What does the convergence of AI model performance mean for businesses? A: Performance gaps between premium and budget AI models are shrinking rapidly. DeepSeek R1 scores 93% versus OpenAI's 95% on benchmarks, meaning businesses can achieve similar results with lower cost alternatives, fundamentally changing procurement decisions.

Q: How quickly is AI moving from digital to physical world applications? A: Extremely rapidly. Tesla accumulated 4,000 million self-driving miles (100x increase in 33 months), Waymo captured 27% of San Francisco rideshare market in 20 months, and xAI built a 200,000 GPU facility in just 7 months.

Q: What does the data suggest about AI's impact on employment? A: AI is creating new job categories faster than it's eliminating others. AI-related job postings increased 448% over seven years while non-AI IT jobs declined only 9%. Companies like Apple have over 600 generative AI openings, suggesting transformation rather than elimination of work.

The Bottom Line

Mary Meeker's 2025 AI report reveals that artificial intelligence adoption is happening faster and more broadly than any technology transition in history. The combination of improving performance, falling costs, and massive capital investment is creating unprecedented opportunities and challenges.

The data suggests we're still in the early stages of AI transformation. As Meeker concludes, "Only time will tell which side of the money-making equation the current AI aspirants will land," but the scale and speed of change indicate AI will fundamentally reshape how business operates across every sector.

For organisations, the message is clear: AI adoption is not optional, it's a competitive necessity. The companies and countries that move fastest to integrate AI capabilities effectively will capture disproportionate value in the emerging AI powered economy.

---

Talk to G3NR8 about AI and your organisation - we help companies like Samsung, JP Morgan and Visa to create empowered workforces and businesses through AI training, strategy and implementation.