Amazon’s $775 million Kiva buy anchors Bain’s physical AI push

Amazon paid $775 million for Kiva Systems eight years after Bain Capital Ventures led its first funding round; Bain said the initial investment generated a 26.6-fold return. Agarwal backed Kiva when Silicon Valley had largely turned away. He now sees a broader opportunity in physical AI, with cheaper hardware, stronger models and deeper technical talent drawing far more capital.

Why did Bain back Kiva when Silicon Valley would not?

In 2004, Ajay Agarwal met Kiva cofounder Mick Mountz. No revenue. No finished product. No customers. Silicon Valley investors had largely looked elsewhere, while software offered a quicker, cheaper route to market.

Mountz proposed a different model: instead of making warehouse workers walk aisles to retrieve products, robots would bring the shelves to them. Agarwal said: “This is the best meeting I’ve been in in 18 months. I didn’t know anything about robotics or hardware, but he explained the problem in a way that you could explain it to your grandmother.”

Convincing Bain’s partners took more work because their initial reaction was that they knew nothing about robots. Through Bain’s network, Agarwal connected Kiva with Don Ralph, who ran North American e-commerce fulfillment for Staples, for additional due diligence. A one-hour meeting became three.

The next day, Ralph, who had 35 years in the industry, called Agarwal, described Kiva as the single most compelling technology he had ever seen and urged him to invest. Staples became Kiva’s first customer; Zappos, Gap and Zara followed.

Eight years after Bain led the first funding round, Amazon bought Kiva, and Bain later reported the 26.6-fold return on its initial investment.

The takeaway: Bain invested before the market, and Amazon’s acquisition supplied the payoff.

Why does Agarwal expect physical AI funding to keep rising?

Robotics and physical AI startups raised $27.6 billion across 1,009 deals in 2025, more than double the previous year’s total, according to PitchBook. That surge reverses the capital pattern behind Kiva’s early backing.

Agarwal said: “We’re in the golden age of hardware and physical AI.” He defined the field as technology that lets robots and other machines perceive and act in the real world. This time, the rest of Silicon Valley has joined.

Bain’s latest early position shows how the market has changed. The firm first backed construction-robotics startup TerraFirma in 2024. In July, TerraFirma announced a $100 million Series A led by Kleiner Perkins, with Bain investing again.

Agarwal’s case rests on three forces:

• Talent. TerraFirma founders Noah Schochet and Noah McGuinness came from SpaceX. Physical Intelligence, another leading robotics startup, began under Berkeley roboticist Sergey Levine and former Google DeepMind researcher Karol Hausman. Agarwal sees that talent pool as one reason the momentum can last.
• Cheaper hardware. Many crucial components now cost a fraction of earlier levels. Smartphone and electric vehicle makers buy cameras, sensors and batteries in massive quantities, and that scale lowers prices for robot makers. Agarwal says lower costs make each project’s return on investment much more compelling.
• Better AI. More advanced models give machines a better chance of handling real-world situations they have not seen before. Agarwal says they can adapt and adjust to the edge cases that inevitably arise in physical environments.

Agarwal also attributes renewed founder interest to 20 years of workflow software leaving some people jaded. They may feel that software is not ambitious enough, he says, and want to build something bigger and harder.

The takeaway: Cheaper inputs, stronger models and technical talent underpin the new case for physical AI.

Can Agarwal find another uncrowded wager?

Twenty-two years ago, Agarwal had to persuade Bain’s partners to take robots seriously. Today, his stated challenge is finding a bet that the rest of Silicon Valley has not already crowded into.

The original wager carried the risk of an unknown category and widespread investor rejection. The new risk lies in a field that attracted $27.6 billion across 1,009 deals in 2025 and drew the rest of Silicon Valley. Bain’s next test is to find an overlooked application before imitation closes the opportunity.

The takeaway: Physical AI no longer needs proof of viability, but the contrarian edge now requires an uncrowded opportunity.