Martin Mao and the trouble with getting comfortable

On the day Uber went public, Martin Mao was already discussing another business. Jerry Chen, the Greylock investor, remembers meeting him to talk about the plan that would become Chronosphere. An IPO is a reasonable occasion to pause and admire the scenery. Mao had brought a question about what came next.

That small episode helps explain a career spent moving toward work he had yet to master. He had worked at Microsoft, Amazon Web Services and Uber. He knew how to build infrastructure. Starting a company would require him to learn why a customer signs a contract, how a sales team works and what happens when the person expected to have the answers has some fairly elementary questions.

In May 2019, Mao and Rob Skillington founded Chronosphere. Six years later, Palo Alto Networks announced an agreement to buy it for $3.35 billion. The transaction closed on January 29, 2026, and Mao became senior vice president and general manager of observability. Those are the large numbers and the formal titles. The more revealing parts of his story concern the work that came before anyone could put a price on it.

The friend who needed a third teammate

Mao grew up in Sydney, and his route into programming began with a vacancy on a team. His school friend Prashant Varanasi needed a third person for a coding competition and taught him enough Visual Basic to participate. They did well enough to win university scholarships. A friend had needed help; a profession followed. It is a pleasingly practical beginning for someone who would spend years making complicated systems useful.

After university came Google and Microsoft. At Microsoft, Mao worked on the transition from desktop Office to Office 365. A product once delivered on discs was becoming a service. His later career would keep returning to that transition: software moving into environments that changed continuously, with engineers responsible for keeping it running.

His own 2020 conference slides put his AWS years at 2013 to 2015. There, he was a technical lead on EC2 and AWS Systems Manager. These were lessons in operating software for other people at considerable scale. The job gave him experience with the machinery behind a cloud service, where a successful launch also creates a continuing obligation to the people using it.

Then Skillington helped draw him to Uber. Their relationship had begun at Microsoft and would outlast several job changes. By the time they became co-founders, there was much more between them than a shared enthusiasm for databases. Skillington had officiated at Mao’s wedding. Choosing a business partner after that much acquaintance leaves fewer surprises about the person, even if the business provides plenty.

A database with a day job

Mao joined Uber in 2015. The observability team faced a problem familiar in outline and punishing in volume: how to understand what a fast-growing collection of services was doing. Each service produced measurements. Each measurement could carry details about location, software version or other characteristics. The combinations multiplied, and storing and querying them became expensive engineering work.

At Uber, Mao and Skillington helped create M3, the open-source metrics platform. Greylock’s original investment account describes a system handling tens of billions of time series. Behind those measurements were practical questions about a business carrying passengers: what was working, what had slowed down and where engineers should look first. A measurement that arrives after the useful moment has passed is an unusually elaborate souvenir.

The founders’ opportunity grew as container-based architectures spread. More companies were assembling applications from smaller services and encountering some of the problems Uber had already faced. Mao saw a chance to sell the product his team had once wanted to buy. Experience inside one demanding company might become useful outside it.

Open source supplied a substantial starting point. In a January 2025 podcast conversation, Mao discussed how years of work on M3 at Uber gave Chronosphere a head start. But the commercial offering needed to solve a wider problem than hosting a database. Customers wanted monitoring they could use, with someone responsible for operating it. The database had to earn its keep inside a service.

Learning to be the person who asks

The technical founders divided the work. Skillington remained focused on technology as CTO; Mao took the CEO role and the business side. The arrangement sounds tidy when written in a sentence. In practice, it meant one experienced engineer deliberately spending his days on work for which engineering experience offered only partial preparation.

Before raising money, they wrote an 80-page business plan. It covered the product, market, competitors and how to reach customers. The document helped them decide whether they believed their own argument. Eighty pages is a generous amount of space in which to discover a bad assumption. Their first audience was themselves.

Chen and Greylock backed the company early. Mao has also recalled a handshake over the first financing at Portland’s Multnomah Whiskey Library. The setting gives the origin story a welcome change of furniture. Infrastructure companies require databases, contracts and patient technical explanations; apparently there was room for a bar as well.

Capital did not remove the need to persuade customers. In a 2025 interview, Mao described repeatedly contacting executives on LinkedIn and asking for conversations. Existing relationships helped with the first customers, but a young company still had to earn permission to replace essential software. Buyers needed a reason strong enough to justify taking the risk.

He also sought advice from people further along: sales and marketing advisers, founders and CEOs introduced through investors. “We stumbled our way in,” he said of enterprise sales. The admission is useful because it preserves the difference between understanding a technical problem and knowing how to build a business around it. He had to learn both.

“We stumbled our way in.”

Martin Mao, on learning enterprise sales

The office beside the washing machine

One of Chronosphere’s early offices was Mao’s laundry room. In his account of the company’s second birthday, he described it as the room from which $55 million in financing had been raised. A laundry room has few of the usual attributes of a venture-backed headquarters. It does, however, offer a fairly direct reminder that ordinary life continues while someone is explaining a very large market.

By May 2021, the company had more than 60 employees across more than 30 cities. Its distributed structure was already part of the operation. Hiring and building a team did not depend on everybody arriving at one address. Mao’s own base remained the Seattle area.

The product became generally available in January 2021 alongside a $43.4 million Series B announcement. Later that year came a $200 million Series C. Speaking about the new funding, Mao described a familiar imbalance: revenue had grown much faster than the team. The company needed people across engineering and the commercial operation to keep up with demand.

His public biography supplies a modest counterweight to the financing figures. He enjoys soccer and meat pies. Those details do not explain a valuation, and they needn’t. They help return the person to human proportions after a sequence of funding announcements. A founder can spend the week talking about distributed systems and still have perfectly ordinary ideas about lunch.

The bill is part of the engineering

Mao’s argument about observability repeatedly comes back to cost. Modern software generates a great deal of information about itself. Keeping it all creates a bill; making sense of it creates work. Chronosphere’s proposition was to give customers more control over the data they paid to retain and the signals they actually needed.

In conversation with Corey Quinn, Mao pushed beyond the search for a universal spending benchmark. The more useful question, he argued, was how observability costs changed as a business grew. A bill that looks manageable today can become a problem if it rises faster than the infrastructure or activity it supports. The direction matters as much as the current total.

He has also argued for open standards such as Prometheus and OpenTelemetry. In a conversation with Heather Joslyn, he welcomed customers having more freedom to move their data between tools. For a vendor, that position carries an obligation: keep earning the customer’s choice. A convenient exit makes the quality of the service rather harder to avoid discussing.

By September 2025, Chronosphere reported more than $160 million in annual recurring revenue. In November, Palo Alto Networks announced the acquisition agreement. The announced consideration covered the company, in cash and replacement equity awards. It was a new corporate setting for a business built around the stubborn expense of understanding what software is doing.

Changing his mind about the chat box

The acquisition closed in January 2026 with a plan to connect Chronosphere’s observability capabilities to Palo Alto Networks’ Cortex AgentiX. Mao’s remit now sat inside a security company, with an ambition to connect finding an operational problem to taking action on it. The work continued beyond the transaction announcement.

One of the more interesting developments is how his own thinking moved. In a May 2025 interview, Mao questioned whether chat was the right starting point for observability. An engineer responding to an alert might naturally want graphs and a visual investigation. A conversational box alone did not settle the question of how useful the software would be.

On October 1, 2026, announcing Cortex XCOR, he described a different experience. As the agents improved, he found the conversational interface becoming his main way into the product. His account emphasized systems that could investigate and explain their reasoning, supported by operational context. It was a change in preference tied to what he said he had seen the software do.

There is a satisfying continuity here. The engineer who became a CEO by learning unfamiliar work was still willing to revise his answer about how engineers should work. An acquisition had supplied a larger organization and a new title. It had also left him with something he had been looking for since the next business plan interrupted IPO day: a problem substantial enough to keep learning.