The Day I Realized Everything Changed
December 2022. I’m staring at my screen in disbelief.
ChatGPT just crossed one million users.
In five days.
Instagram took 75 days to hit one million users. Twitter took two years. Netflix took three and a half.
Five days broke every adoption curve I knew.
That’s when it hit me: this wasn’t just another platform shift. It was something faster—and messier. Adoption without permission.
The world changed quietly. Most companies were already behind.
The $37 Billion Land Grab
In 2016, AI was a sub-$1B market. By 2025, it hit roughly $37B. By 2026, it’s expected to pass $45B.
The interesting part isn’t the growth. It’s who moved.
Traditional banks spent months evaluating AI-driven lending. During that time, Upstart deployed models that approved 44% more borrowers at similar default rates. Affirm captured about 30% of the buy-now-pay-later market—roughly $37B in annual payment volume.
Same tools. Different urgency.
One group studied. The other shipped.
Three Questions That Decide Outcomes
Are You Winning or Losing?
GE Healthcare has 72 FDA-cleared AI devices. Siemens has 47. Philips has 38.
GE holds roughly 25.5% global market share.
Not because they had better algorithms.
Because they moved earlier and kept moving.
At the same time, startups like Aidoc and Viz.ai didn’t try to “do AI.” They solved one painful problem—stroke detection, emergency imaging—and shipped fast.
Hospitals that committed early signed multi-year contracts. The rest are now paying premium prices for basic capability.
Waiting didn’t reduce risk. It just changed who captured the value.
Where Will You Get AI Talent?
At Snap, I ran a research team of 50+ people working on AI systems used by 188 million daily users.
When we hired, we competed directly with Google and Amazon.
Compensation mattered—but it wasn’t decisive.
Top researchers routinely accepted $20,000 less to trade scale for ownership—choosing smaller, problem-focused teams over incremental optimization roles inside Big Tech.
Later, as CTO at Heali AI, I couldn’t match Big Tech pay. What we offered instead was ownership: building an AI system from scratch with direct impact on health outcomes.
That attracted the right people.
The AI talent war isn’t about salaries. It’s about whether your problem is worth committing years of your life to.
Most companies choose one of four paths: build, buy, partner, or contract. All have tradeoffs.
But doing nothing is still a decision—and it’s the one that guarantees you fall behind.
What Could Go Wrong?
IP leaks In April 2023, Samsung disclosed three incidents in 20 days where engineers pasted confidential data into ChatGPT—including semiconductor source code.
Cyberhaven estimates that 3.1% of employees using AI tools submit sensitive data. In a 100,000-person company, that’s hundreds of leaks per week.
Regulatory risk In July 2025, Massachusetts reached a $2.5M settlement with Earnest Operations over AI lending discrimination. The model penalized applicants based on college default rates. No one coded bias. The system learned it from historical data.
It still broke the law.
Reputation risk The Apple Card controversy showed how fast things escalate. A 2024 Lehigh study later confirmed the issue: identical AI-generated loan applications received different outcomes based on demographic signals inferred from data.
The question isn’t “does it work?”
It’s “can you explain it when things go wrong?”
Three Moments That Changed Everything
1997: Deep Blue beats Kasparov AI crushes problems with fixed rules.
2017: AlphaGo teaches itself AI finds strategies humans never would.
2022: ChatGPT goes mainstream Millions adopted it overnight. No training. No approval.
I’ve talked to executives who discovered their teams had been using AI in production for months.
One told me: “We spent six months evaluating tools. Our developers had already been using ChatGPT for four.”
Your team is already using AI.
The only question is whether you know where—and how.
Ask Better Questions
Most teams focus on the wrong things: model size, benchmarks, architecture.
The questions that matter are simpler:
How do we know when the system is wrong?
What happens when it fails?
Can we switch vendors if we need to?
What data trained it—and what bias came with it?
Good questions surface problems early. Bad questions give false confidence.
What’s Coming Next
The next wave of AI won’t just generate text.
World models understand physics, causality, and space. DeepMind’s Genie 2 can generate interactive 3D environments from a single image. Think robotics, manufacturing, logistics, automation.
And then there’s quantum computing. If it matures faster than expected, it could compress years of AI progress into months—and break today’s encryption.
You don’t need to understand the math.
You need to understand the exposure.
The Window Is Closing
The $37B AI market will look small in hindsight.
Every quarter of delay compounds the gap between companies that shipped and companies that waited.
We’re still early—but not early enough to sit this out.
After twenty years in AI, one thing is consistent:
The winners aren’t the ones with the fanciest models.
They’re the ones who move early, learn fast, and ask the right questions.
The shift is already happening.
The only question is whether you’re ahead of it—or reacting to it.
Dr. William Brendel has spent over 20 years building and leading AI systems across research, product, and healthcare, helping companies turn emerging technology into real-world advantage.
