Apoorva Mehta tried 20 ideas. Then he opened the fridge.

Apoorva Mehta’s first Instacart delivery had a pleasingly compact chain of command. He placed the order. He went to the grocery store. He brought the groceries home. Customer satisfaction was, presumably, within reach. On June 2, 2012, a future grocery business began with an engineer delivering dinner to himself.

The scene is more revealing than the later valuations. Here was a service with almost none of the machinery that makes a company look like a company. Yet its central transaction already worked. Someone wanted groceries, an order reached a phone, and a person fetched them. The organizational chart could wait until after checkout.

For Mehta, this modest arrangement followed roughly 20 unsuccessful attempts at building products. He had spent two years learning how many things software could do without discovering something enough people wanted it to do. His refrigerator proved a more useful adviser than his earlier brainstorms. It had little to offer except hot sauce and a recurring complaint.

An excellent app for nobody

Among the attempts was a social network for lawyers. It had the outward ingredients of a startup: a proposed audience, a product, a team, and the prospect of a professional community. What it lacked was a founder with much firsthand knowledge of the people he hoped to serve. Mehta had never even worked with a lawyer.

He persisted with the idea for about a year. Eventually, he and his collaborator parted ways. The experience exposed an uncomfortable distinction between enjoying the act of making software and caring about the task it performed. A founder can be busy for months while the customer remains entirely untroubled by the product’s absence.

Mehta’s later account of those failures contains less swagger than the familiar Silicon Valley celebration of persistence. Each attempt taught him something; each also depleted his confidence. He acknowledged that a few more failures might have exhausted his willingness to continue. Twenty is an arresting number. Living through them was evidently less decorative.

A grocery order for one

His interest in groceries had a longer history than that San Francisco kitchen. Born in Jodhpur, India, in 1986, Mehta moved to Libya as a baby, then to Canada with his family in 2000. Hamilton, Ontario, became home. Later, he remembered waiting at a bus stop near Toronto, carrying shopping bags in the freezing cold.

He studied electrical engineering at the University of Waterloo, graduating in 2008. His early experience included BlackBerry and Qualcomm. At Amazon, he worked on supply-chain systems, encountering the practical mathematics of getting orders from warehouses to customers. Software was connected to physical consequences: a package, a route, a delivery that had to happen.

In 2010, he left Amazon and moved to San Francisco. Entrepreneurship initially meant trying one possibility after another. Grocery shopping eventually supplied something more durable than novelty. He knew the inconvenience himself. By 2012, ordering goods through a screen was familiar; assembling a week’s groceries still involved an expedition. He began coding a way around it.

“I wanted to make grocery shopping effortless.”

Apoorva Mehta, 2018

The important response came after his own test order. Friends started using Instacart without his having to persuade them. That was a new sensation after the previous products. There is an excellent kind of customer feedback that arrives without an adjective: another order. Mehta had found a problem that continued to exist when he stopped talking about it.

The application arrived with beer

He wanted to join Y Combinator, but the application deadline had passed two months earlier. He sought introductions to its partners and kept approaching people who might listen. Garry Tan offered a narrow opening: he could submit a late application, though admission at that point would be difficult. Mehta treated the remaining possibility as usable space.

He sent Tan a six-pack of beer through Instacart. It was a demonstration with the advantage of being drinkable. Rather than asking a partner to imagine an on-demand transaction, Mehta arranged one. A meeting followed, and Instacart entered Y Combinator’s summer 2012 batch. The deadline remained a deadline; the functioning product had supplied another reason to talk.

Max Mullen and Brandon Leonardo joined as co-founders during the company’s early development. The origin is often told as a solo performance because the first order makes such a tidy anecdote. The growing business required other people, other judgments, and the much less tidy work of building a team. A good founding story leaves room for its supporting cast to become essential.

The supermarket was already built

Online groceries came with an awkward inheritance. Webvan’s collapse had made the category a cautionary tale. Mehta’s answer rested on what had changed: connected smartphones could coordinate customers and shoppers, while existing supermarkets already held the inventory. Instacart could organize access to those stores without first reproducing the entire supply chain.

That model made local selection part of the proposition. In an early interview, Mehta explained that Amazon Fresh had not given him access to everything he wanted, including Trader Joe’s. Customers had preferences that did not disappear when they opened an app. The grocery store was a collection of particular choices, rather than an interchangeable building full of food.

Getting those choices into the app could be comically physical. When the team could not obtain a store’s product catalog online, it bought items, took them to a studio, photographed them, and entered the information. The digital shelf began with actual groceries. One imagines a particularly philosophical checkout assistant confronting that shopping cart.

Mehta also fulfilled early customer orders himself when necessary, sometimes using Uber because he did not own a car. By 2013, the company was studying how weather affected demand and delivery. Rain helped inform its interest in Chicago. The service sold convenience, but providing it required taking inconvenience seriously, down to the journey through a particular neighborhood.

The details grew more exacting as the business expanded. Shopper software helped locate items and handle replacements when something was unavailable. In 2017, Mehta described fixes to bottle deposits and sales-tax calculations that improved order economics by cents. Grocery delivery could attract grand pronouncements; it still had to account for the bottle.

When the old employer bought a partner

Amazon returned to the story in 2017 by agreeing to buy Whole Foods, an important Instacart partner. The overlap was uncomfortable: a company that had helped teach Mehta fulfillment was acquiring a retailer his own company served. Other grocers had a clearer reason to consider an alternative delivery partner, but the immediate uncertainty was considerable.

The Whole Foods relationship eventually wound down. In December 2018, Instacart announced a transition affecting its in-store shoppers, with the first group of 243 affected in February 2019. The company outlined efforts to relocate workers and provide separation support. Strategic changes appear as arrows in presentations. For the people picking orders, they change where the next shift happens.

Mehta had to manage that less charming side of a marketplace, too. Changes to shopper compensation attracted criticism. Customers, retailers, shoppers, and advertisers brought different interests to the same system. Keeping an order moving was an engineering problem; keeping those relationships workable demanded decisions no routing algorithm could make pleasant for everyone.

A founder learns to leave the chair

On August 2, 2021, Fidji Simo became Instacart’s CEO and Mehta moved to executive chairman. He had invited her onto the board earlier that year. Her experience running Facebook’s main app brought a different scale of operating knowledge. Mehta publicly backed her appointment while acknowledging how difficult it was to imagine someone else running his life’s work.

The following year, the company announced he would leave the board when it became public. That happened in September 2023. His departure closed a particular chapter of responsibility, even as the business he helped establish continued. The first order had needed him at every step. The public company had people whose jobs were to take it further.

The change fits something Mehta has said about his preferences. Developing a business model and putting its pieces together gave him pleasure. As a company grew, the founder’s work shifted toward allocating resources, selecting problems to address, and representing the organization. The title could remain familiar while the daily occupation became something quite different.

Another change in the equation

There are clues to his range of interests on his personal website. A reading list, explicitly dated to 2016, moves among scientific history, invention, decision-making, and competitive strategy. It includes How We Got to Now and The Everything Store. This is a bookshelf with room for both the origins of an idea and the trouble of operating it.

By April 2026, another venture was public: Abundance, an investing firm started the previous year. It had raised $100 million in seed equity financing and was using AI agents for research and trading, with some strategies operating entirely through AI. Its stated ambition was to extend that role across the investment process. The company’s claims will have to meet the market’s accounting.

Mehta’s subsequent hiring message sought someone to lead technical investing for his personal portfolio. He wanted deep investigation of technology alongside commercial judgment. The emphasis is recognizable from Instacart: understand what a new capability permits, then examine the business it might support. An earlier outcome can inform that work without settling its result.

“Honestly, I just didn’t want my legacy to be defined by just one company,” he said in a conversation with investor Elad Gil. The next attempt now has a different setting, and considerably more capital. Still, his most useful early distinction survives: wanting a company is easy. Finding work that holds your attention after the pitch is over takes rather more shopping around.