From Google Labs to AI Unicorns: Why Bradley Horowitz’s Seed-Stage Bets Are Worth Watching
When you’ve led product at Google, backed OpenAI before it was a household name, and written checks to Anthropic, Cerebras, and Slack before they hit scale, people pay attention to what you do next.
Bradley Horowitz, the former Google executive who oversaw product management for Google Labs, Photos, and Streams, has quietly built one of the most impressive early-stage portfolios in tech through his venture firm, Wisdom Ventures. He wrote early checks to OpenAI in 2023 and Anthropic in 2024—two bets that now define the generative AI boom. His portfolio also includes Cerebras (custom AI chips), Slack (acquired by Salesforce for $27.7 billion), ScaleAI (data labeling for AI), and dozens of other high-growth companies.
That track record has landed Horowitz—once again—on Business Insider’s Seed 100 list, a curated ranking of top seed-stage investors. But the question every founder, operator, and VC is asking now is simple: Where is he putting capital to work today?
We sat down with Horowitz (via a recent interview with senior correspondent Ben Bergman) to unpack his investment thesis, founder criteria, and the signals he’s watching for the next wave of breakout companies.
The Wisdom Ventures Thesis: Why Horowitz Backs Founders Before Products
Horowitz doesn’t invest in categories. He invests in founders.
That might sound like a cliché in venture capital, but his track record suggests there’s real discipline behind the mantra. When he wrote the check to OpenAI in 2023, the company was already generating massive buzz, but it was still pre-revenue on many of its current revenue streams. When he backed Anthropic in 2024, the competitive landscape was already crowded with foundation model players like Cohere, Mistral, and Google’s own DeepMind.
So what does Horowitz see that others miss?
“I look for founders who have a clear vision of the future—not just a better mousetrap,” Horowitz said in the interview. “They understand the infrastructure changes that are coming, not just the applications.”
That lens explains his bet on Cerebras, the chip company building wafer-scale processors for AI workloads. Most VCs were focused on software layers. Horowitz bet on the silicon layer, predicting that model training and inference would become compute-constrained far sooner than the market expected. He was right.
What Horowitz Looks For in Seed-Stage Founders
Horowitz has a reputation for being a high-conviction, low-friction investor. He writes small checks early, rarely leads rounds, and gives founders operational leverage through his network and pattern recognition from Google.
Here are the three signals he screens for:
1. Deep technical insight + market timing
Horowitz wants founders who understand a technology shift before it becomes obvious. He backed ScaleAI because he saw that every AI company would eventually need high-quality labeled data—and that most teams were underestimating how hard that would be at scale.
2. Founder-market fit that isn’t generic
Horowitz doesn’t invest in “AI for X.” He invests in founders who have lived inside the problem for years. That might be a former Google Brain researcher building a new model architecture (Anthropic) or a fintech veteran rethinking compliance infrastructure.
3. Capital efficiency with ambition
He’s skeptical of founders who raise massive rounds before proving product-market fit. “The best companies find a way to get to $10 million in ARR with $5 million or less,” he noted. “If you need $20 million to find product-market fit, you’re doing something wrong.”
The Sectors Horowitz Is Watching Now
Based on recent portfolio moves and his public commentary, here are the areas Horowitz believes will produce the next generation of breakout companies:
AI Infrastructure (Not Just Applications)
Horowitz’s bet on Cerebras reflects a conviction that the AI infrastructure layer—chips, networking, data centers, and orchestration software—will generate more durable value than most application-layer AI companies. As models commoditize, the underlying hardware and middleware become the moat.
Data Moats for Enterprise AI
He’s interested in companies that own proprietary, hard-to-replicate datasets. ScaleAI fits this thesis, but so do startups building domain-specific data pipelines for healthcare, legal, and manufacturing. “The companies that own the data will own the value chain,” Horowitz said.
Developer Tools for the AI Era
Horowitz sees a gap in tooling for AI engineers. Most existing DevOps and MLOps tools were built for traditional software. He’s backing startups that reimagine monitoring, debugging, and deployment for AI-native workflows.
Vertical SaaS with AI-Native Features
He’s not chasing horizontal AI “copilots.” Instead, he’s interested in vertical SaaS companies that use AI to dramatically reduce time-to-value in specific industries—legal contract review, medical coding, supply chain optimization.
Lessons for Founders: What Horowitz’s Track Record Teaches Us
If you’re a seed-stage founder, here’s what you should internalize from Horowitz’s approach:
Don’t optimize for the warm intro. Optimize for clarity of vision.
Horowitz responds to founders who can articulate a future that differs from the present in specific, testable ways. If you pitch him “we’re like [hot company] but for [vertical],” you’ve already lost.
Beware the “AI wrapper” trap.
He’s skeptical of startups that are simply wrapping GPT-4 or Claude with a UI and calling it a product. “If your moat is a prompt template, you don’t have a moat,” he said.
Think about the infrastructure shift, not just the app.
The biggest winners of the past decade (AWS, Stripe, Twilio) were infrastructure bets that powered thousands of applications. Horowitz believes the same pattern will repeat with AI—and he’s betting accordingly.
Why Horowitz Keeps Making the Seed 100 List
Business Insider’s Seed 100 isn’t a popularity contest. It’s based on deal volume, portfolio performance, and founder feedback. Horowitz’s repeat appearance reflects a consistent ability to identify early-stage outliers before they become obvious.
His portfolio is not the largest. It’s not the most aggressive. But it has one of the highest hit rates in the industry—and that’s a direct result of his conviction-driven, founder-first approach.
“I’d rather write 10 checks to exceptional founders than 100 checks to average ones,” Horowitz said. “The math works out better.”
The Takeaway for Growth Teams and Revenue Leaders
If you’re in B2B SaaS or tech, here’s why this matters to you:
The companies Horowitz backs often become your future customers, partners, or competitors. Understanding his thesis gives you a preview of which verticals, technologies, and business models will be funded—and which will struggle to raise.
- Watch the infrastructure layer. If Cerebras, ScaleAI, and similar companies succeed, your go-to-market motion will need to account for AI-native procurement and deployment cycles.
- Prepare for vertical AI disruption. If Horowitz is right about AI-native vertical SaaS, many traditional SaaS companies will face new, more efficient competitors within 18 months.
- Invest in proprietary data. Companies that own unique, structured data will have pricing power. Those that don’t will be margin-compressed.
Final Thoughts
Bradley Horowitz may not be the loudest investor in the room. But when he writes a check, it’s worth paying attention.
His bets on OpenAI, Anthropic, Cerebras, ScaleAI, and Slack weren’t random. They were the product of a disciplined, thesis-driven approach that filters for exceptional founders at the exact right moment in a technology cycle.
For founders, operators, and investors: study his pattern. And if you get the chance to pitch him, make sure your vision of the future is clear, contrarian, and backed by real technical depth.
This article is based on reporting by Ben Bergman for Business Insider, Bradley Horowitz’s inclusion on the Seed 100 list, and public information about Wisdom Ventures’ portfolio.