The 300 page AI report
The 4 top take-aways Mary Meeker's 2025 AI Trends, the most comprehensive AI report of the year.
If you’ve been in tech for more than a decade, you remember the ritual: Mary Meeker’s annual “Internet Trends” report would drop, and the industry would stop to listen.
Now, Meeker (Queen of the Internet) and her team at BOND have done it again, but this time for a force moving at a speed that makes the early internet look like it was stuck in traffic.
Their new report, "Trends – Artificial Intelligence," isn't just a data dump; it's a stunning narrative of a world being fundamentally and irrevocably reshaped, not in decades, but in months. [Report]
In 1999, Vint Cerf said a year online was like a dog year. Today, the internet's pace feels slow compared to AI’s. And this time, machines are driving the acceleration.
Forget slow burn. AI isn’t just moving fast — it’s folding time. From user growth and developer adoption to enterprise spend and geopolitical pressure, AI is rewriting how information flows, how companies compete, and how the world works.
“The pace and scope of change related to AI is indeed unprecedented.” – Mary Meeker, BOND Capital
The report’s central, unmissable message is one of unprecedented velocity. The rules, players, and power dynamics of technology are being rewritten in what feels like real-time.
This isn't just another tech cycle; it's an industrial revolution on fast-forward, fueled by a tsunami of capital and geopolitical urgency.
Here’s the story the BOND report tells, and what it means for everyone.
1. The New Speed of Light: "We Have Not Seen Likes of Before"
The most arresting insight from the report is the sheer, vertigo-inducing speed of AI adoption. Past technological shifts had "dog years." AI operates on an entirely different clock.
As BOND puts it, this is “Growth We Have Not Seen Likes of Before.”
ChatGPT vs. The World: It took Netflix over 10 years to reach 100M users. It took ChatGPT two months. The report uses this as a foundational indicator that the playbook for tech adoption has been torn up. AI isn't just faster; it’s a categorically different phenomenon.
Global from Day One: The internet revolution, as BOND notes, "started in the USA and steadily diffused globally." AI, in contrast, "hit the world stage all at once." In just its third year, the ChatGPT app already sees 90% of its users outside of North America—a level of global penetration that took the internet 23 years to achieve. This is a profound shift. There is no longer a single epicenter of user growth; the revolution is decentralized and immediate.
2. The $212B Arms Race: "Inflected With AI's Rise"
This new reality is being built on a mountain of cash. The report highlights the astronomical capital expenditure (CapEx) from the "Big Six" U.S. tech companies (Apple, NVIDIA, Microsoft, Alphabet, Amazon, Meta).
The big number: These giants spent a collective $212B on CapEx in 2024, a staggering 63% increase year-over-year.
What they're buying: This isn't just for more servers. As BOND explains, this is a strategic pivot "toward computation / intelligence." They are building the factories of the 21st century—vast, power-hungry AI data centers. Microsoft's Brad Smith is quoted saying, "Like electricity and other general-purpose technologies in the past, AI and cloud datacenters represent the next stage of industrialization."
The Profit Paradox: Here’s the tension at the heart of the AI boom. While revenue for AI companies is skyrocketing, so are the costs. BOND’s chart on a leading AI company’s revenue versus its compute expense is a masterpiece of storytelling. It shows a glorious, upward-sloping revenue line and a terrifying, downward-plunging expense line (representing massive losses).
3. The Key Players: A Shifting Battlefield
The report paints a picture of a multi-front war with four distinct armies.
The Incumbent Titans (Microsoft, Google, Meta, Amazon): Their weapon is distribution. With billions of users across search, social, cloud, and operating systems, they can roll out AI features to a global audience overnight. They are integrating AI into everything—from Microsoft Copilot in Office to Google’s AI Overviews—to defend their moats and create new revenue streams.
(Source: BOND) The AI Darlings (OpenAI, Anthropic, xAI, etc.): These are the challengers defining the frontier. They are in a capital-intensive race to build the most powerful models. Their challenge, as the report points out, is monetization. They have the buzz and the tech, but the incumbents have the existing customer relationships and cash flow.
(Source: BOND) The Kingmaker (NVIDIA): In this gold rush, one company is selling all the picks and shovels. NVIDIA is the undisputed hardware king. The report shows its data center revenue has exploded, now capturing a stunning and rising share of all global data center CapEx. Everyone, from the incumbents to the darlings, is a customer.
The Open-Source Rebellion (Meta's Llama, Hugging Face): This is the wild card. Open-source models are rapidly closing the performance gap with their closed-source rivals, but at a fraction of the cost. This, BOND notes, is "leveling the playing field for AI development" and threatening the pricing power of the closed-model giants.
4. The New Geopolitics: "Our Space Race"
BOND makes it clear: this is not just a commercial competition. It’s a geopolitical struggle with the intensity of a new Cold War.
USA vs. China: The report quotes Meta’s CTO Andrew Bosworth, who describes the current state of AI as "our space race and the people we’re discussing, especially China, are highly capable..." The data backs this up. China is not just a fast-follower; it's a leader in key areas. It has already surpassed the rest of the world in the installation of industrial robots and is rapidly catching up in large-scale AI model development.
A Bifurcated World: China is successfully fostering domestic champions like DeepSeek, which are gaining massive traction locally while Western platforms are blocked. This is creating a splintered AI internet, where platform choice is "increasingly shaped by national identity."
So what?
The BOND report isn’t just tracking trends. It’s documenting a full-blown reset.
As Mary Meeker and team show, we’re not in an era of tech evolution — we’re in a moment of tech upheaval.
Here’s what it means:
1. AI is laying the groundwork for a new economy
The buildout is massive — data centers, GPUs, sovereign models. It’s not just software anymore. It’s infrastructure.
As the report bluntly states: “There is no AI without energy.”
2. Revenue is booming — but so are the losses
AI leaders are growing fast but burning faster. Billions in compute costs. Monetization lags.
The model feels familiar: Amazon, Tesla, Netflix — all lost heavily before dominance. But not everyone makes it through.
3. The real battle isn’t product — it’s distribution and infrastructure
Big Tech can scale AI instantly to billions through search, email, and productivity tools.
But the real power lies deeper — with those who control compute and energy. Think NVIDIA. Think utilities.
4. A new class of competitors is rising
The next breakout company won’t look like Google. It could be two developers using open-source models like Llama or Cursor, building something niche, fast, and profitable.
Barriers to entry are gone. Distribution is king. But infrastructure is God.
5. This is not just about technology — it’s about global power
U.S. vs. China isn’t theoretical. It’s active. Whoever leads in AI, leads in influence.
The stakes are national, not just corporate.
What’s next: Three immediate plays for leaders
The AI curve is steep, and waiting is riskier than acting. Based on the BOND report, here are three non-negotiable plays every leader should be making now:
Mandate AI fluency, not just an "AI strategy." Top-down strategies won’t cut it in a world moving this fast. Winning teams are those where AI isn’t a project — it’s a reflex.
Think Shopify, Duolingo: employees use AI daily, by default.
Train your teams to work with AI tools, question outputs, and build faster. Fluency beats formality.Interrogate your infrastructure stack. Your biggest tech decision now: OpenAI or open-source?
The performance gap is shrinking.
Closed models offer speed; open ones offer control and cost advantages. Leaders must pick a lane — and invest accordingly. This is your new build vs. buy moment.Chase workflows, not novelty. AI’s future isn’t in conversation. It’s in doing.
Agents that schedule, design, code, analyze — end-to-end.
The real value? Automating workflows your business relies on daily. Find one. Own it. Scale it. That’s where durable competitive edge — and monetization — lives.
In short: Educate. Evaluate. Execute.
AI is not a department. It’s a capability. Make it everyone’s job — before your competitor does.
The report ends on a clear note:
“It’s gametime for AI, and the genie is not going back in the bottle.”
The old playbooks are out. What worked in mobile or cloud won’t cut it here.
This is a new cycle — one of infrastructure, speed, and endurance.
Those who adapt fast — and build deep — will define the next decade.
Best,
Marc & Team