The AI Tsunami
Mary Meeker is back. And her 340-page AI report is the roadmap every leader needs.
Remember Mary Meeker? You Should.
For over a decade, Mary Meeker’s Internet Trends reports were required reading for anyone trying to understand where the digital world was heading. They were dense, data-rich, forward-looking, and often eerily accurate. Then she stepped back. And the industry felt it.
But now, she’s back—with Bond Capital’s “Trends in AI” report: a 340-slide masterclass on the state of artificial intelligence in 2025.
And make no mistake: this is not just another AI hype deck. This is a benchmark report—one that lays out the stakes, winners, losers, and inflection points for what might be the most transformative technology since the internet itself.
1. Unprecedented Adoption
ChatGPT didn’t just break the internet. It reset the standard. 800 million weekly active users in just 17 months, dwarfing adoption curves of the iPhone, Instagram, or TikTok.
What that tells us: AI isn’t a phase. It’s an operating system shift.
2. Costs Are Collapsing—But the Model Is Flipping
Training models is still wildly expensive—some costs reaching nine figures.
Inference (running the models) has dropped ~99% since 2020, opening the door for smaller players.
It’s like the early cloud days: huge upfront costs, but radically cheap to scale once the infrastructure is built.
3. Hardware Is the Hidden Hero
The NVIDIA Blackwell chip uses 105,000× less energy per token than a 2014 GPU. That’s not just Moore’s Law—it’s fuel for the next industrial revolution.
Meanwhile, Google (TPUs), Amazon (Trainium), and other hyperscalers are in an all-out race to own the AI compute stack. Because whoever controls the chips controls the game.
4. Global Momentum & Open Source Velocity
China has emerged as a heavyweight in open-source AI. U.S. players like Meta and Mistral are pushing the boundaries of open models. It’s no longer West vs. East—it’s open vs. closed, cloud vs. edge, speed vs. scale.
The competitive heat map is red-hot. And this time, the margins aren’t just financial—they’re geopolitical.
5. Regulation, Ethics & the Emerging Guardrails
Bond’s report doesn’t shy from the uncomfortable: deepfakes, AI bias, workforce displacement, data sovereignty. These are not side effects. They are central battles.
Meeker and team argue: governance needs to be proactive, not reactive. Because once models are widely deployed, it’s hard to rewind the consequences.
What It Means for You (and Your Business)
📈 Adoption & Strategy: Embed AI not as an add-on but as a core competency.
💸 Cost Structure: Plan for high training costs, but design systems optimized for low-cost deployment.
⚡ Infrastructure: Choose chips & providers based on energy efficiency, not just raw power.
🌍 Competition: Be China's rise-aware; leverage open-source tools to stay agile.
⚖️ Ethics & Governance: Build your AI frameworks now—before someone else forces them on you.
Mary Meeker helped define the internet era. With this report, she’s doing it again for AI.
Bond’s Trends in AI isn’t just a snapshot—it’s a blueprint. A clear-eyed, data-rich, brutally honest map of where we are and what’s next.
So if you lead, build, hire, invest, or create—don’t just skim it. Digest it. Debate it. Use it.