Peter Lipka and the price of a millisecond

Peter Lipka’s LinkedIn introduction consists of five words: “Delivery above everything else.” It is a compact calling card for someone whose working life has involved an unusually expansive idea: that computers should support worlds large enough for crowds to inhabit together. The ambition has acquired different names over the years. Lipka’s job has repeatedly involved getting the machinery, and the organisation around it, to work.

He is now chief executive of Somnia, a blockchain project whose current interests include decentralised finance, prediction markets and onchain credit. Before that came Improbable, where he was a co-founder and chief operating officer. Further back lies Goldman Sachs. This looks like a tour through several technology fashions until you notice a recurring practical concern: how quickly a complicated system can respond when a great many things happen at once.

The pixels before the platform

At Imperial College London, Lipka built a final-year AI project that learned to play a video game from pixels on a screen. He graduated in 2011 with a first-class MEng in Computing. The project makes a neat opening scene for his career: software observing a world, discovering its rules and doing something inside it.

Goldman Sachs followed. The move into finance placed him in another environment where the underlying system’s performance mattered. Describing the connection later, he put it plainly: “High-frequency trading is basically ‘speed is money.’” He saw techniques from that work as useful to the software being built at Improbable.

He joined the venture in 2012, alongside Herman Narula and Rob Whitehead. Asked about their ambition, Lipka recalled finding the pair rather wild and the technical challenge attractive. His account contains a short, cheerful verdict: “I loved it.” It is a better explanation for taking a difficult job than a paragraph of management vocabulary.

A company grows around the code

Lipka helped shape SpatialOS, the distributed operating system at the centre of Improbable’s early work. He also led its fabric, engineering and commercial teams before becoming COO. In the company’s first five years, its headcount grew from four to nearly 200. His responsibilities expanded along with it.

The sequence matters. An engineer can spend a long time concentrating on whether a system works. Commercial leadership adds the question of who needs it. Operations brings in the people and arrangements required to deliver it repeatedly. Lipka’s route passed through all three. His technical contribution remained part of the record, too: he is named among the inventors on a patent application titled Simulation Systems and Methods.

In May 2017, Improbable received $502 million in a funding round led by SoftBank. Horizons Ventures and Andreessen Horowitz also participated. For the young company, it was a substantial change in resources and expectations. A large investment can pay for more attempts at a hard problem. It also gives far more people a reason to ask how those attempts are going.

Improbable’s subsequent history moved through multiplayer technology, virtual events and venture creation. Its current description of that evolution emphasises spinning specialised teams into independent companies, then incubating new businesses with capital and practical support. That is a considerable distance from simply selling one piece of software. The accumulated engineering experience became something the organisation could apply in several places.

When the operating model changes

By September 2023, Lipka was describing a leaner Improbable and an intention to initiate, grow and eventually spin off businesses exploring the metaverse. The company’s activities covered technology development, building worlds and experiences, and building ventures. It was also working with partners including Major League Baseball. The range was broad; the operating task was to give each activity a workable place.

The financial picture was changing as well. Improbable reported £78 million in revenue for 2022 and a £19 million loss, a reduction of £131 million in losses. Those figures gave the restructuring a concrete backdrop. An imaginative business still has a cash account, and it tends to be less susceptible to imaginative interpretation than the product roadmap.

For 2023, Improbable reported its first full-year profit, £11 million, alongside £66 million in revenue and £185 million in year-end cash. The company credited growth, cost discipline and asset disposals. That qualification belongs with the number: a profitable year involving divestments tells a more particular story than a simple claim that the original business had found an effortless formula.

Lipka’s public explanation of venture building focused on helping founders obtain capital and providing a foundation from which to operate. It was a recognisable extension of his earlier work, bringing technical ambition into contact with the arrangements that allow a team to pursue it. The product had expanded to include the conditions in which other products could be built.

The T-shirt test

His argument for connected virtual worlds becomes easier to grasp when reduced to a piece of merchandise. In a 2025 interview, he discussed a concert T-shirt: in physical life, it can leave the venue with you; in a restricted virtual world, its usefulness may end at the door. Wider use, he argued, gives the digital item more value. The example has the virtue of requiring neither a headset nor a glossary.

MSquared’s proposition involved independent virtual worlds connected through shared infrastructure. Lipka advocated an open network and independent businesses rather than control concentrated in one platform. The T-shirt was a modest way into a large question about who sets the rules for digital possessions. Even an imaginary wardrobe can acquire a surprisingly officious landlord.

There was engineering behind that argument. Improbable described a demonstration in which a Bored Ape avatar appeared in MLB’s Virtual Ballpark, involving ownership verification and permissions across environments. The details explain why moving a digital object is more involved than copying its appearance. Someone has to establish what it is, who can use it and how the receiving world should treat it.

“Delivery above everything else.”

Peter Lipka’s LinkedIn introduction

Finding another use for speed

Somnia brought a related infrastructure problem into view. In October 2024, Improbable announced development of a layer-one blockchain designed for mass consumer applications. The proposed uses included gaming and social platforms. Compatibility with the Ethereum Virtual Machine was intended to let developers bring existing skills and applications to the new network.

The announcement also described a specialised database, ICEdb, a custom EVM compiler and compression techniques. These are the sorts of components that disappear behind the interface when a product feels responsive. Lipka, then Improbable’s COO, presented Somnia as an example of the projects its venture-building operation could support with engineering talent, technology and resources.

In February 2025, Improbable published results from Somnia’s November devnet tests across 100 nodes in multiple regions. The reported throughput varied by task: 1.05 million ERC-20 transfers per second, 300,000 NFT mints and 50,000 Uniswap trades. Each workload asks different things of the system. Putting the figures side by side is more informative than allowing the largest one to stand for everything.

Benchmarks describe capability under specified conditions. Their commercial significance depends on what developers build and what people choose to use. That gap between capacity and demand is familiar territory for an executive who has worked across both engineering and commerce. It remains a live question rather than something a fast test can settle.

A different seat, a changing market

Somnia announced Lipka as CEO on 23 March 2026. Paul Thomas continued as founder, concentrating on the network’s long-term technical and strategic vision. Harry Lang joined as chief marketing officer and Kevin Zia as chief operating officer. Zia had been chief of staff at Improbable, bringing another connection to the organisation from which much of the infrastructure experience came.

The arrangement put Lipka in charge of execution alongside a founder retaining a defined strategic role. It also required a distinction between related organisations. Somnia’s foundation is legally and operationally independent, although it was incepted by MSquared. Lipka’s new title belongs to Somnia; his Improbable experience explains part of what he brings to it.

By September, Somnia was describing a shift begun earlier in 2026 toward DeFi, prediction markets and applications built around onchain agents. Its stated next priorities included onchain credit. The network’s original work for games and metaverse applications was being directed toward another set of users. The infrastructure had continued while its intended commercial centre of gravity moved.

The same update said investors, team members, launch partners and advisers had extended their token lockups by nine months, moving the first unlock to June 2027. Lipka’s stated competitive ambition was direct: “We are building with one goal in mind: to make it irrational to compete”. The next evidence will come from products and their users.

For Lipka, the circle back toward financial infrastructure is striking. He left a bank for a difficult computing problem; years later, the technology around him is serving markets again. The intervening work has involved code, teams, capital and several changes of direction. Five words still sit at the entrance to his professional profile. The useful part of his story is what delivering them has required.