Sergey Brin, using Google Glass as an example, reminds founders not to release products too early. They should first refine core performance and the differentiated value, avoiding falling into the speed obsession and misaligned iteration.
As a founder, this is the most sincere + practical startup sharing I’ve seen in the past year: don’t do lean startup.
Recently, I’ve mainly been working on two things:
The most impactful startup principle I’ve encountered—and hope to always adhere to—is from a recent public talk by a well-known entrepreneur:
“Don’t release your product too early.”
This was Sergey Brin’s response at a campus event at Stanford in December 2025.
The background is that in December 2025, during the centennial celebration of Stanford Engineering School, Sergey Brin, co-founder of Google, was invited back to campus for a dialogue. The participants were Stanford President Jonathan Levin and Dean of the School of Engineering Jennifer Widom.
A student asked how to avoid pitfalls when starting a business:
Brin’s core advice: don’t make a big splash and launch before the product is ready—he directly cited Google Glass as an example. The main idea is: if you have a cool new hardware idea, you must first thoroughly refine it before doing flashy launch events like skydiving or flying in a blimp.
This sharing was incredibly sincere. Most entrepreneurs at such events tend to share politically correct views or some high-level motivational clichés that sound inspiring but, in reality, lack concrete action steps. But Brin offered a very practical perspective.
We spent a lot of time, made many mistakes, and paid significant costs to realize the importance of this advice.
Because the startup philosophies we’ve been following—like lean startup, lightning-fast iteration, user-first, rapid cycles—may be flawed.
Let’s first examine Brin’s core reasoning: because once you release too early, it’s hard to tell whether you’re on the right path of iteration or just patching user desires. Once you start signaling externally, it’s like being on a “treadmill”—you’re bound to a delivery schedule, but you may not have enough time to complete everything you should. Meanwhile, external expectations snowball, growing larger and larger, while you lack the time to digest, judge, and manage these expectations.
From my personal entrepreneurial experience, another key reason is that releasing too early might mean you haven’t yet considered two critical questions:
Take our project UniversalX as an example—we “perfectly” made these two mistakes:
We didn’t realize that there was still an opportunity for product-driven growth in the market (or even evaluate this possibility). We placed too much emphasis on the so-called timing window. Fundamentally, we were overly opportunity-driven, and the underlying issue was a systemic laziness—more about chance than strategy.
Since we didn’t assess whether there was still a product-driven opportunity, we couldn’t make optimal decisions about the core support needed for a product-driven approach. Our differentiator—later discredited as “multi-chain”—was based on this flawed premise. But the market proved that for trading terminals, product-driven growth can only rely on: information advantage (alpha—at least making users feel they have alpha) or time advantage (performance).
It wasn’t until Axiom launched—relying on product performance and, despite being late (in what seemed like a highly competitive, red ocean environment)—that we understood about 80%. Not 100%, because we continued to make mistakes afterward, not going all-in on alpha and performance, but still trying to align and fill in features. We are still paying the cost for this today—yes, we’re spending more time aligning performance, even though it’s been a year since we launched, and 90% of people already believe the trading terminal industry is pointless.
In short: we started our business too easily equating “speed + iteration” with an unbreakable truth, neglecting to think about where the real decisive step in market competition lies. We also too readily took early user feedback as positive reinforcement, which often leads to misaligned iteration directions and increases the sunk costs of subsequent adjustments (time + emotional).
In the AI era, this is even more true. Tools have leveled the productivity gap and strengthened information equality. This drastically reduces the cost of producing “just good enough” products and products without leverage in design, making the term “crash-and-iterate startup” meaningless.
As the saying goes: when lamps are everywhere, what you wish for becomes more important.
Stop doing lean startup. Stop doing lightning-fast startup. Think carefully about what your true product aspirations are.