Things are About to Get Weird: The March of AI

If you thought things were moving fast last year, buckle up. The transformation I described in “Ten Minutes on a Sofa” wasn’t a blip, it was a warning shot for how radically and weirdly our technology landscape is about to tilt. The pace of AI innovation is now so wild, it’s warping the rest of the software market in its wake and the march forward isn’t just about software anymore. Robotics, intelligence, even the whole idea of what companies buy and sell, is mutating in real time.​

SaaS M&A Collapse, AI M&A Explosion

2025 is the year the old playbooks broke. Traditional SaaS businesses are finding exits harder than ever, with M&A deal value plummeting. Q1 saw enterprise SaaS M&A drop nearly 25% quarter-over-quarter, and the mega-round drought is real: deals over $100 million cratered to just 21 in the past year, a dramatic collapse from 147 just a few years ago. If you’re building “classic SaaS” without AI fuelling outsized growth, you’re competing for a shrinking pool of capital in a changed world.​

Meanwhile, the AI industry is flooded with investment and deal activity. AI M&A broke records in 2024, over $200 billion in deals globally, up 20% year-on-year and 2025’s numbers so far show another 21% rise in deal volume in Q1 alone. It’s not just about volume. Average deal size in AI has doubled year-on-year, and megadeals (over $1B) are now the norm. In the first half of 2025, AI accounted for over 50% of global venture capital spend, dwarfing SaaS.​

If SaaS was king, AI has stolen the crown and melted it down for parts.

Leaders Predict Models 1000x More Intelligent

It isn’t hype. The founders of Anthropic, OpenAI, and other AGI developers are publicly forecasting a leap to models 1000x more capable than today’s “base” agents, possibly by 2027 or 2028. Benjamin Mann (Anthropic) suggests a 50% chance of hitting some form of superintelligence by 2028. Industry modelling expects a “software-driven intelligence explosion” where the leap from human-level to super-intelligent systems compresses decades or centuries of R&D into months, driven not by more hardware, but by better algorithms.​

This acceleration isn’t linear. AI is now learning how to improve itself, automating both coding and research, fuelling cascading leaps that are unpredictable even to experts. If it took me ten minutes on a sofa this week to ship an agent improvement, what happens when next year’s models are 50x, 200x, 1000x smarter?

Robotics: The Silent Revolution Is Roaring

Away from the screen, robots are being transformed by AI. In 2025, modular robots that assemble lunar bases (see NASA ARMADAS), repair undersea pipelines, and operate with a precision no human can match (0.02mm) in autonomous surgery, are already a reality. Investment in robotics platform companies is surging (over $4.2 billion last year). Collaborative robots ‘cobots’ now work alongside humans in small business, adapt in real time to changing conditions, and can be programmed by non-experts.​

Key shifts:

• AI-driven autonomy: robots that analyse, plan, and adapt on the fly.

• Collaborative innovation: tech giants and startups unite to invent adaptable machines.

• Advanced sensing: haptic sensors, computer vision beyond human accuracy.

From manufacturing to healthcare to space, robots are no longer dumb machines, they’re becoming teammates, explorers, and, in time, self-improving innovators.

Speculation: Where Next? How Fast?

What does this mean for the next few years? Wild speculation, but here’s what the signs suggest:

• The gap between human imagination and implementation collapses: models will create, test, and re-engineer entire software ecosystems in minutes, not months.

• Traditional approaches (like SaaS compounding ARR) will be redefined, or replaced, by AI-native platforms and autonomous agents that iterate hundreds of times faster.

• Robotics will extend beyond the factory floor, radically re-shaping logistics, infrastructure, and even creative work.

• The market will polarize: companies not embracing AGI and adaptive robotics risk obsolescence.

• The rate of change is exponential, not incremental, meaning that “weird” is the only thing you can expect.

If last years progress felt rapid, this year is moving at lightspeed. Problems that seem impossible today could be solved in weeks. But with this pace comes risk and responsibility. The same tools that supercharge progress could cause huge harm if not handled with care.​

It’s a weird, exciting, precarious time. The question isn’t whether things will get weird, but whether we’re ready for the velocity of the change. Are you building AI responsibly? Are you building for safety, alignment, and real human benefit, or for speed alone? As the transformation unfolds, those who get it right will be the ones who shape our future.

Weird is just beginning. Let’s make sure we build it well.