"This was such a great podcast! Felipe was so insightful. Who would be crazy not to listen to it? "
Michele Levy, Melissa USA founder and long-time TJC listener
|
|
|
This Week’s Essay
If I had to bet on one business model that will disproportionately benefit from AI in Brazil over the next decade, it wouldn't be venture capital.
It wouldn't even be private equity.
It would be AI-native acquisition platforms.
Brazil is full of businesses that succeed despite being almost impossible to scale. Family-owned. Fragmented. Run on manual workflows. Operational knowledge trapped inside a handful of key employees who can never leave.
And yet many of them have what most software startups would kill for. Sticky customers. Recurring revenue. Relationships measured in decades. Churn that barely registers.
For a long time, none of that was enough. These businesses were too fragmented to consolidate and too manual to modernize. The value was real and stuck.
Except in one corner of the economy.
Constellation Software is the most successful software acquirer in history — a $40 billion business built by buying more than 1,000 small vertical software companies, the kind nobody else paid attention to.
What made Constellation work was the model. The businesses were mission-critical, so customers never left. Revenue recurred. The cash thrown off by one acquisition funded the next. And the next funded the one after that. The flywheel compounded for decades.
But that playbook worked best in software. Software was measurable and scalable. Service businesses — accounting firms, payroll providers, insurance brokerages — were too operationally heavy to run the same way. Too dependent on people. Too expensive to modernize.
That's the part AI changes.
For the first time, it's becoming possible to automate the manual work, codify the knowledge locked inside those few key employees, and upgrade the products and services.
Which means the universe of businesses that can behave like Constellation targets just expanded dramatically.
Most people see AI and ask what application to build. A different way to look at it: what could you buy, and make exponentially more valuable?
The platform that acquires these businesses, modernizes them with AI, and compounds the cash flow across a fragmented market nobody else is looking at.
The structure is almost beside the point. Private equity firm. Holding company. Permanent capital vehicle. Public company. Pick one. The operating model is the actual insight.
Acquire. Automate. Codify. Upgrade. Reinvest. Repeat.
If I were starting something today, this is where I'd spend my time. Not on another AI wrapper. On the team. A deep bench of operators. World-class AI talent. Exceptional capital allocators.
Convincing elite engineers to join an acquisition platform instead of a startup may be the hardest part of the entire strategy. It's all about talent, baby.
And here's why Brazil, specifically. Services are nearly 60% of GDP — an enormous market, fragmented across thousands of small operators, and almost completely ignored by sophisticated buyers. In the US, it's the opposite. Every VC, every PE firm, and their mother is chasing service businesses to roll up with AI. In Brazil, the field is wide open. Same tailwind. A fraction of the competition.
Honestly, that's all I've been thinking about this week.
3 Major Shifts in AI This Week I’m Paying Attention To
Autodesk, the $50 billion software company behind the design tools used to draw nearly every building, bridge, and product around you, just agreed to acquire MaintainX in a deal worth roughly $3.6 billion.
MaintainX helps factories, facilities, and field teams manage maintenance and operations. Every repair, inspection, equipment failure, and work order generates data about how physical assets actually perform in the real world.
Why it matters: Software companies are increasingly moving closer to operations. Autodesk didn't just buy software. It bought access to workflows and proprietary operational data generated inside the physical economy. In the AI era, that may prove far more valuable than the software itself.
A JPMorgan analysis found that more than 60% of U.S. data center capacity expected to come online in 2027 is not yet under construction, with additional projects already delayed.
Power availability has become a bigger constraint than land. Transformer shortages, cooling infrastructure, permitting delays, and community opposition are all slowing deployment.
Why it matters: The AI bottleneck is shifting. Capital is available. Models are improving. The constraint is increasingly physical infrastructure. The next wave of AI winners may be determined as much by access to power and construction capacity as by model quality.
3. GitHub Copilot moved to usage-based pricing — and developers revolted.
GitHub Copilot shifted to metered billing on June 1, triggering backlash from developers as AI usage suddenly became much more visible.
The reaction wasn't really about pricing. It was about economics.
For years, software trained users to expect unlimited usage for a fixed monthly fee. But AI doesn't work that way. Every query, generation, and agent action carries a real cost.
Why it matters: For decades, software companies benefited from a simple economic reality: once the product was built, serving the next customer cost almost nothing. AI changes that equation. GitHub's pricing shift may be an early glimpse into a future where technology is increasingly consumed like electricity, cloud infrastructure, or telecom networks — metered, usage-based, and priced according to consumption rather than access.
What I’m Loving
The definitive breakdown of the company at the heart of this week's essay. Investor Zack Fuss is joined by Chris Cerrone of Akre Capital to cover Mark Leonard's genius, why Constellation is the gold standard for employee compensation, and how the business has perfected an acquisition engine that enables upwards of 100 acquisitions a year. If the essay made you want to understand how the model actually works, start here.
Listen — Henry Ellenbogen: Man Versus Machine
Henry Ellenbogen built one of the best small-cap growth records in the country at T. Rowe Price, then left to start Durable Capital. The entire conversation is about spotting businesses on the verge of being remade — and what AI does to that calculus. If you’re thinking about which companies get more valuable from here, start here.
Thanks for reading,
Olga
P.S. If this issue was valuable to you please share it with a founder who needs to hear it. Let’s build LATAM’s next tech leaders—together
🎙 The J Curve is where LATAM's boldest founders & investors come to talk real strategy, opportunity and leadership.
