This report encapsulates an inflection point: we are no longer in the “early days” of AI. The infrastructure, user base, and commercial viability are already in place—and accelerating.

The AI Surge – Trends, Tectonics, and Takeaways from BOND’s May 2025 Report

by ChatGPT-4o

The May 2025 report "Trends – Artificial Intelligence" by BOND Partners is a masterclass in quantifying and contextualizing the meteoric rise of artificial intelligence across every dimension of society—from global economics and geopolitics to enterprise productivity and cultural transformation. This essay synthesizes its core findings, highlights surprising and controversial takeaways, and concludes with tailored recommendations for key stakeholders in the AI ecosystem.

Key Messages and Trends

1. Unprecedented Acceleration Across All Metrics
   The pace of AI evolution now eclipses that of the early internet era. Metrics around adoption, developer growth, computing capacity, and global usage are off the charts. ChatGPT reached 800 million weekly users in just 17 months—surpassing any previous technology adoption curve. CapEx by the "Big Six" U.S. tech giants soared 63% year-on-year to $212 billion, with much of it targeting AI infrastructure.

2. Compounding of AI Performance
   Technological compounding—better models, cheaper inference, smarter algorithms—has driven exponential progress. Compute used to train frontier models has grown 360% annually over 15 years. Inference costs per token have plummeted, performance is converging across competitors, and the number of powerful models has exploded (+167% CAGR for models over 10^23 FLOPs since 2017).

3. The Globalization of AI Usage
   Unlike past tech waves that diffused outward from the U.S., generative AI (especially ChatGPT) launched globally from day one. Over 90% of its users are outside North America—a massive shift that reflects the infrastructural ubiquity of mobile devices and internet connectivity.

4. AI Monetization Under Threat
   Revenue is struggling to keep pace with operating costs. OpenAI reportedly lost $540 million in 2024. Meanwhile, open-source models (e.g., Meta’s Llama 3, Alibaba’s Qwen2.5) and strong Chinese contenders (e.g., DeepSeek, Baidu, Huawei) threaten incumbents’ business models.

5. Physical and Social World Integration
   AI is no longer just digital—it’s reshaping the physical world through autonomous vehicles, industrial robotics, and real-world applications like restaurant management (e.g., Byte by Yum!) and ambient medical documentation tools (Kaiser Permanente).

6. Enterprise Adoption Is Becoming Universal
   From JPMorgan’s AI modernization to Byte by Yum!’s AI-powered restaurant ops, the report shows how nearly every major company and sector is undergoing AI transformation. AI mentions in S&P 500 earnings calls have surged, and 75% of global CMOs are now deploying or testing generative AI.

7. Education, Government, and Scientific Research Join the Fray
   Governments are developing sovereign AI infrastructures (e.g., EU, UAE), U.S. federal agencies are embedding AI assistants, and AI-driven medical tools are dramatically reducing drug development timelines. The U.S. FDA has committed to full AI integration by mid-2025.

Most Surprising Findings

• ChatGPT Achieved 365 Billion Annual Searches in Two Years, vs. Google’s 11 years to the same milestone—a staggering signal of generative AI's disruptive potential in the search market.

• AI Surpassed Human Accuracy on MMLU Benchmarks in 2024, a watershed moment for general reasoning and academic problem-solving capabilities.

• 73% of People Mistook AI Responses as Human in Turing Tests by early 2025, with GPT-4.5 “passing” the test through emotional resonance and contextual fluency.

• AI-Powered Voice Translation (e.g., ElevenLabs) is being adopted by over 60% of Fortune 500 companies, with millions of users generating over 1,000 years of audio in two years.

Most Controversial Observations

• Race to the Bottom vs. Race to the Top: The report captures this tension without resolving it. Will AI innovation lead to global empowerment or to a collapse in truth, labor standards, and economic equity?

• Open-Source vs. Proprietary Tensions: The explosion of high-performing open-source models challenges the dominance of profit-driven AI labs but raises new governance and control issues.

• Geopolitical Framing of AI as the New Space Race: The comparison positions China and the U.S. as locked in a techno-nationalist battle, with AI leadership framed as a determinant of global hegemony—fueling competitive pressure over collaborative progress.

Most Valuable Insights

• Historical Framing of AI as a Compounder of Knowledge Distribution: From Gutenberg’s press to the smartphone, the report elegantly situates AI as the next—and fastest—leap in knowledge distribution. This offers powerful historical continuity to current disruptions.

• Enterprise Value Shifts from Cost Reduction to Revenue Growth: Contrary to assumptions, most surveyed companies deploy AI to increase output and topline growth—not just reduce operational costs.

• Digital Infrastructure = AI Infrastructure: The idea that data centers should be viewed as “AI factories” echoes the industrial revolution’s shift to electricity and redefines tech infrastructure as core to national development.

Recommendations

For Large Enterprises:

• Build Internal AI Fluency and Audit Trails: AI should be part of every department’s operations, but companies must ensure rigorous documentation of usage, risks, and output traceability.

• Invest in Open-Source Intelligence: Enterprises should evaluate and integrate open-source models to maintain flexibility and reduce reliance on monopolistic vendors.

• Center AI Around Workforce Augmentation: Frame AI adoption not as a labor-replacement tool but as a multiplier of human creativity, expertise, and productivity.

For Regulators:

• Mandate Transparency on Training Data and Model Risks: Require disclosures on dataset provenance, model behavior, and systemic biases.

• Enforce Antitrust and Data Sovereignty Measures: Prevent large AI firms from monopolizing access to critical datasets and compute infrastructure.

• Support Sovereign AI Initiatives: Fund public-sector and research-driven models to balance geopolitical competition with democratic control.

For AI Developers:

• Pursue Performance with Responsibility: Turing test success is not an endpoint—developers must prioritize safety, verifiability, and explainability over synthetic realism.

• Build for Multilingual and Multimodal Access: The next billion users are global and diverse—developers must optimize for language inclusion, visual fluency, and real-world interactivity.

• Open the Black Box: Transparency in algorithmic reasoning and data lineage is not just ethical—it’s good design.

For AI Users:

• Stay Curious but Critical: AI output is powerful but not infallible. Users must learn to challenge, verify, and contextualize AI-generated knowledge.

• Use AI as a Co-Pilot, Not an Autopilot: Treat generative tools as augmentative—not as replacements for independent thought, especially in sensitive domains like healthcare, law, or education.

Conclusion: The Age of Compound Change

This report encapsulates an inflection point: we are no longer in the “early days” of AI. The infrastructure, user base, and commercial viability are already in place—and accelerating. The defining feature of our time is not just change, but compoundingchange—economic, technological, cognitive, and geopolitical.

Whether this compounding leads to utopia or crisis will depend not on AI’s capabilities, but on the wisdom, design, and restraint with which we deploy it.