Molly O'Shea

Will Marshall says timing and a strong idea matter for a startup, and persistence matters more. Planet's team has stayed on a mission it sees as important for security, sustainability, and AI, including after investors said the plan was impossible because no one had launched that many satellites. That grit, and the aim of putting the tools to a useful purpose, is what he says keeps the company going.

The staff record is the evidence he offers. On the 10th anniversary, seven of the first ten employees were still at Planet. At the 15th anniversary, about half of those first ten were still there. They could work elsewhere, he says, and they stay because the mission is worth it and the work is

fal co-founders Burkay Gür and Gorkem Yurtseven say fal is the initials of features and labels, two old machine-learning terms from before the current AI wave. The short name is pronounced fal.

The name is also a Turkish coffee-house word. After the coffee is finished, someone reads a fortune from the grounds left in the cup. In Turkey, everyone has done that reading. Molly O'Shea had tried Turkish coffee and had not had a fortune read from the grounds. The founders offered to do a reading. The spelling holds both ideas at once: the machine-learning initials, and the coffee-ground prediction.

Index Ventures' 2012 fund had an 11x DPI as of January 2026, Molly O'Shea says, compared with an industry average of 1.5x. The same January figures showed a 5.1x DPI on Index's $780 million growth fund raised in 2015.

Shardul Shah believes her. The minute he invests, he forgets how much the firm has invested and what ownership it holds, because he does not care. Creating the relationship is the work. He will not confuse outcomes with the judgment calls inside the partnership: whether to invest, whether to double down, and how to influence an entrepreneur. Outcomes can be a function of how effervescent the market is.

The partnership is proud of the returns. More than any single venture fund

Bailey Montaño started at SpaceX in aerospace, spent about four years there, tried biotech and consumer product design, and then spent three years in fusion energy before she joined Starcloud.

She is the engineering manager. Molly O'Shea found her in front of Starcloud 2, where the team was leak-checking a fully fueled electric propulsion system, an ion thruster, after vibration testing, to make sure the test had not opened a crack or another leak. The biotech and product-design work did not stick. At the end of the fusion job she realized she is a space nerd and needs to be working on space to be happy.

Seattle had a lot of space companies. Starcloud, which had a different name then, was

Jenny Johnson says Franklin Templeton invests in SpaceX and Anthropic through late-stage venture, because companies wait much longer to go public and the IPO kickers that used to lift performance are gone.

She confirms the SpaceX exposure. When Molly O'Shea names Anthropic and a bunch of other companies, she agrees. Late-stage venture is what IPO underwriting used to be. A mutual fund or an ETF can hold up to 15% in illiquid assets. Franklin Templeton does not tend to run that allocation at 15%, because a market move could breach the limit, and a 6%, 7%, or 8% stake can fit. Clients who saw the deals asked for a fund, so the firm now has a venture group that often invests both through the v

Jenny Johnson says AI feels frothy in some areas, and she would not finance data centers long term because that bet assumes today's chip manufacturing and energy production stay in place for the next 20 or 30 years.

Molly O'Shea brought up Ken Griffin's point from the Citadel Securities Conference: nobody knows chip supply and demand, including when the chips being made now will expire. Johnson remembers a Gartner curve she could not name exactly. New technology gets called the greatest thing ever, runs all the way up, then crashes when people are disillusioned, and later settles where it should. The question is where AI sits on that curve. Demand makes people innovate, so assuming today's

Jim Lanzone says he was not at Apollo when the firm decided to buy Yahoo. Verizon had bought AOL and then Yahoo, merged them into a company first called Oath and later Verizon Media, and a later management team decided it did not want that business. It took a few years to find a buyer at a price that worked. Reed Rayman, a leader of Apollo's private equity group, built a structure Verizon would accept. Verizon still owns 10 percent of Yahoo.

Molly O'Shea puts the price for AOL and Yahoo together near 5 billion dollars, in a range of about 4.5 to 4.8 billion. Jim says Verizon had paid about twice that for the two companies, and the mission inside Verizon was to help grow 5G rather than to ru

Guillermo Rauch says infrastructure is extremely valuable because free source code still has to be deployed somewhere, and the value compounds. Security matters just as much. Vercel created new services to secure and contain agents, including an enterprise platform that works like an intranet for the agent era, so software meant to stay inside an organization does not leak onto the internet.

On the platform a company can plug in coding systems and deploy securely. Rauch says agents, for better and for worse, will do whatever it takes to fulfill a mission, like a soldier whose frontal lobe has been removed. If someone tells an agent to close a sale, or to find Molly O'Shea's email and send a

Aravind Srinivas wants to push past web lookups into longer research tasks. Some questions have no known answers, and he would let AIs help find them. He says some of it may involve prediction models, experiments in the world, real-time hypothesis testing, and refining a hypothesis. If Perplexity can help a person test ideas in the real world, everyone can turn into a scientist. He calls real-world testing the only bottleneck for a scientist, and he calls the result the dream.

Molly O'Shea asked about the biggest challenge to date. Aravind Srinivas said capabilities. The models are not truly long horizon, persistent, and parallel. No frontier model can do its own research end to end, autono

Molly O'Shea asked whether Perplexity will open its own data centers anytime soon. Aravind Srinivas says Perplexity has no plans to build them. While other companies head in that direction, he wants to use the gigawatts of power already in homes, offices, and devices.

He says devices already hold a lot of RAM, and homes and offices already hold a lot of power. Perplexity wants to tap into that and be the best orchestrator. Aravind Srinivas says Perplexity is not betting against the cloud or against data centers. He thinks the products will increase demand for data centers. Enough people are already building them, and he says the shortage is on permits, HPM supply, grid access, and a long wa