What the Inventors Share

Ten portraits, one pattern — and a note from the author

By VastBlue Editorial · 2026-03-26 · 17 min read

Series: The Inventors · Episode 11

The Pattern

Over ten episodes, we profiled eleven people who built technology the world depends on. Wayne Westerman, who taught glass to feel because his hands hurt too much to type. JB Straubel, who made electric cars inevitable by treating 6,831 laptop batteries as a single instrument. Federico Faggin, who drew the first microprocessor by hand and signed his initials in the silicon. Matei Zaharia, who looked at MapReduce and thought: why is this writing to disk? The Palantir builders — Gettings, Nassar, Doyle, LeBlanc — who solved the data integration problem that the 9/11 Commission said was the biggest impediment to national security. Sophie Wilson, who designed an instruction set in BBC BASIC that now runs in six billion devices. Jonathan Herman, who helped coordinate thousands of satellites so that the internet could reach places cables never will. Dimitris Lyras, who digitised the oldest industry on earth because his ships were drowning in paper. Karlheinz Brandenburg, who compressed a revolution into three letters because he kept listening to "Tom's Diner" until he could no longer hear the difference. Michael Armbrust, who applied to Berkeley four times and then built the data lakehouse.

These are not the names on magazine covers. Most of them are unknown to the general public. Several have never given a public interview. One etched his initials in silicon precisely because he knew nobody would credit him otherwise. What they share is more instructive than fame.

Strip away the differences in era, domain, and geography, and ten stories converge on the same handful of structural truths. The inventors in this series share a relationship with problems, a relationship with time, and a relationship with institutions that is remarkably consistent — and remarkably different from the way innovation is popularly understood. What follows is an attempt to state those shared truths plainly, draw out the argument they collectively make, and explain why we believe those arguments matter beyond the pages of this series.

Specific Pain, General Solution

Westerman built multitouch because typing caused him physical pain. Straubel built battery management systems because solar car racing in the Australian desert taught him that energy efficiency was everything. Faggin designed the 4004 because Busicom needed a cheaper calculator chip. Zaharia built Spark because MapReduce was too slow for the machine learning workloads he cared about. Brandenburg spent years listening to Suzanne Vega's "Tom's Diner" because he needed a perceptually transparent test signal to validate lossy audio compression. Herman tackled satellite coordination because the physics of orbital mechanics demanded it if broadband was ever going to reach underserved geographies. Lyras digitised vessel management because his family's ships were generating more paper than cargo revenue. In every case, the inventor started with a specific, concrete, often personal problem. The general solution emerged from the specific constraint — never the other way around.

This is worth stating precisely, because the popular mythology of invention works in the opposite direction. The mythology says: visionaries see the future, then build towards it. Westerman did not see the iPhone. He saw his inflamed tendons. Straubel did not see the global transition from internal combustion to electric propulsion. He saw that his solar car kept running out of charge in the desert between Adelaide and Darwin. Faggin did not see the personal computer revolution. He saw that Busicom wanted twelve chips and he thought he could do it in four. The vision came later — sometimes years later, sometimes decades — when the specific solution turned out to be the general one. The iPhone was a consequence of Westerman's pain. The electric vehicle industry was a consequence of Straubel's dead batteries. The personal computer was a consequence of Faggin's cost reduction.

The Palantir builders illustrate a variation of the same pattern at institutional scale. Gettings, Nassar, Doyle, and LeBlanc did not set out to build a company worth tens of billions of dollars. They set out to solve the data integration problem that the intelligence community had failed to solve before September 11, 2001. The constraint was specific: analysts at different agencies held pieces of the same puzzle but had no system capable of assembling those pieces across classification boundaries, data formats, and organisational silos. The solution — a platform for integrating heterogeneous data sources with fine-grained access control — turned out to be the same problem that every large enterprise faces, from pharmaceutical companies managing clinical trial data to banks monitoring transaction flows for fraud. The general emerged from the specific. The platform company emerged from the security problem.

Even Armbrust, whose work is the most architecturally abstract in the series, followed this pattern. He did not set out to build a new category of data infrastructure. He set out to solve the concrete problem that Spark SQL queries were unreliable when multiple writers accessed the same data lake simultaneously. The solution — Delta Lake, a transactional storage layer — turned out to be the missing piece that made data lakehouses viable as a category. The lakehouse was not the plan. The plan was ACID transactions on Parquet files. The lakehouse was what happened when that specific fix met the market's accumulated frustration with the data warehouse–data lake split.

There is a reason this pattern recurs. Specific problems provide what grand visions cannot: a concrete test of whether the solution works. Westerman could test his multitouch surface by typing on it. If his hands hurt less, it worked. If the gestures were recognised reliably, it worked. He did not need market research, user studies, or a focus group. He needed his own hands. Straubel could test his battery architecture by driving his car. If the car made it across the desert, the thermal management worked. If the cells balanced, the BMS worked. Specificity provides the falsification mechanism that grand vision lacks. You can spend a decade pursuing a grand vision without ever discovering that it is wrong. You cannot spend a week pursuing a specific fix without discovering whether it fixes the problem. This is why the pattern recurs: specific pain selects for inventors who actually test their solutions against reality, and reality is the only filter that produces foundational technology.

Compounding Expertise

Faggin designed the silicon gate process at Fairchild, then the 4004 at Intel, then the 8080, then the Z80 at Zilog, then the Synaptics touchpad. Each built on the one before. The silicon gate process gave him the fabrication knowledge to design a processor. The 4004 gave him the architecture knowledge to design a better one. The Z80 gave him the systems knowledge to understand what happened at the interface between chip and user. The Synaptics touchpad was not a career change — it was the logical extension of five decades spent understanding how silicon could sense and respond to the physical world. Straubel raced solar cars at Stanford, then built the Tesla battery architecture, then designed and built the Gigafactory in Nevada, then founded Redwood Materials to recycle the spent batteries his earlier work was producing at scale. Armbrust created Spark SQL, then Delta Lake, then Structured Streaming, then Photon — each layer enabling and depending on the previous one. Wilson designed the BBC Micro's operating system before designing the ARM instruction set before co-founding the company that would license it to the world.

The pattern is not genius moments. It is compounding expertise applied to adjacent problems over decades. The inventors in this series did not jump between unrelated fields. They stayed close to their core competence and expanded outward, one problem at a time. The depth of knowledge they accumulated in their specific domain gave them the ability to see solutions that generalists could not — because the solutions required understanding the domain deeply enough to know which constraints were real and which were merely conventional.

Consider the difference between how Faggin and a hypothetical generalist would approach the touchpad problem in the early 1990s. A generalist might study the touchpad literature, survey existing products, and attempt to improve the user interface. Faggin understood the problem from the silicon up — literally. He knew how capacitance behaved at the junction level because he had fabricated junctions by hand. He knew how signal processing worked at the chip level because he had designed signal processing chips. He knew how the interface between hardware and human perception worked because he had spent three decades watching users interact with the systems his chips powered. The touchpad was not a new problem for Faggin. It was the surface expression of problems he had been solving at progressively higher levels of abstraction for his entire career.

Brandenburg's career shows the same compounding structure. His doctoral work on perceptual audio coding led to the MP3 standard, which led to the AAC standard, which led to decades of work on spatial audio and immersive sound. Each project required and built upon the psychoacoustic models he had developed in the previous one. Herman's trajectory from orbital mechanics researcher to satellite constellation architect to broadband infrastructure designer follows the same deepening arc — each role demanded everything the previous role had taught, plus the next layer of complexity.

Lyras offers perhaps the most instructive example because his domain — maritime shipping — is not one that technology culture considers glamorous or even interesting. But it is precisely Lyras's multi-generational depth in shipping that made his technology company, Ulysses Systems, possible. He understood which operational problems actually cost money, which regulatory requirements actually shaped behaviour, and which data flows actually mattered — because he had grown up inside the industry. A technologist parachuting into maritime from Silicon Valley would have built a different product, almost certainly a worse one, because they would not have known which constraints were real. Lyras knew. His software worked because his shipping knowledge compounded over decades, and his technology knowledge compounded alongside it.

This compounding effect is poorly understood and poorly rewarded. The technology industry celebrates pivots, disruptions, and "10x thinking." The inventors in this series did not pivot. They persisted. They did not disrupt — they accumulated. The Z80 was not a disruption of the 4004. It was its descendant. Delta Lake was not a disruption of Spark SQL. It was its foundation. Redwood Materials was not a disruption of Tesla. It was its necessary consequence. The mythology of Silicon Valley valorises the founder who drops out, pivots three times, and stumbles into a billion-dollar market. The inventors in this series did the opposite. They stayed. They went deeper. They treated their career not as a series of bets but as a single, decades-long investment in understanding one domain thoroughly enough to reshape it.

The inventors in this series did not pivot. They persisted. They did not disrupt — they accumulated. The mythology of the pivot is a poor model for how foundational technology actually gets built.

The implications for how we evaluate engineering talent are significant. If the pattern that produces foundational technology is compounding expertise rather than brilliant disruption, then the hiring practices, incentive structures, and cultural narratives that reward breadth over depth, novelty over persistence, and pivots over deepening arcs are systematically selecting against the kind of engineers who build the things everything else depends on. We do not claim this is always the case. But across ten episodes and eleven inventors, the pattern is consistent enough to be worth stating plainly: depth compounds, and the compound interest on deep expertise is foundational technology.

The European Thread

Four of the inventors profiled in this series are European. Faggin is Italian, trained in physics at the University of Padua and in semiconductor fabrication at SGS-Fairchild in Milan before moving to Silicon Valley. Brandenburg is German, spending his entire career at the Fraunhofer Institute for Integrated Circuits in Erlangen — a government-funded research organisation. Lyras is Greek, building enterprise software for the Mediterranean maritime industry from a base that straddles Athens and London. Wilson is British, working at Acorn Computers in Cambridge and then at ARM, a company that began as a twelve-person spinout from a failed home computer maker.

Their stories challenge the narrative — popular in American technology media and reinforced by decades of venture capital mythology — that foundational technology innovation happens only in Silicon Valley. It does not. The microprocessor was conceptualised and designed by an Italian physicist. The instruction set architecture that powers the majority of the world's mobile devices was designed by a British engineer working in a small office in Cambridge. The audio compression standard that restructured the global music industry was developed at a German government research institute. Enterprise digitisation of global shipping was driven by a Greek shipowner. The inventive capacity is European. The mythology is not.

What Europe has consistently produced is foundational invention — the kind of deep, technically rigorous work that creates entirely new categories. What Europe has consistently lacked is not inventive capacity but commercialisation velocity. The MP3 was invented at Fraunhofer; iTunes was built in Cupertino. ARM was designed in Cambridge; the ARM-based smartphone ecosystem was commercialised by American and Asian companies. The inventions are European. The platform companies that captured most of the economic value from those inventions are not.

This is a structural observation, not a cultural criticism, and it is worth examining the structural factors with some precision. First, capital. European venture capital has historically been smaller, more conservative, and more fragmented than its American counterpart. A European deeptech startup raising a Series B in 2010 faced a fundamentally different capital environment than an equivalent company in Palo Alto. The rounds were smaller, the timelines were shorter, the appetite for pre-revenue infrastructure companies was lower. This has begun to change — European VC investment has grown substantially over the past decade — but the structural gap in growth-stage capital for hardware-intensive and infrastructure-layer companies remains significant.

Second, market size. The European single market is, in practice, not yet single for technology companies. A SaaS company selling to German manufacturers, French utilities, and Italian logistics firms encounters three different regulatory environments, three different procurement cultures, and three different sets of data sovereignty requirements. An American company selling to customers in Texas, New York, and California encounters one. The fragmentation imposes a scaling cost that has no equivalent in the US or China, and it disproportionately affects the kind of enterprise infrastructure companies that the inventors in this series tend to build.

Third, risk appetite. Europe's publicly funded research institutions — Fraunhofer, the Max Planck Institutes, CERN, the university systems of the UK, Germany, France, and the Nordics — produce world-class science and engineering. But the institutional incentive structure often rewards publication over commercialisation, and the career risk of leaving a tenured position to found a startup is perceived differently in Munich than in Mountain View. Brandenburg could have left Fraunhofer to start a company around perceptual audio coding. He did not — and as a result, the MP3 became an open standard that restructured an industry while Fraunhofer collected licensing revenue rather than building the platform. The outcome was arguably better for the world but worse for European technology sovereignty.

The gap between European invention and European commercialisation is not a talent gap. It is a capital gap, a risk-appetite gap, and a market-size gap. Structural changes to European venture capital, single-market regulation, and scale-up infrastructure are slowly, unevenly, beginning to close it. But the closing is slow, and meanwhile the pattern persists: Europe invents the foundation, and someone else builds the house. Understanding this pattern — and understanding that it is structural rather than inevitable — is one of the reasons this series exists, and one of the threads that connects it to our next series.

The Institutional Context

There is another pattern that deserves explicit acknowledgement: every inventor in this series worked within an institutional context that made their invention possible, even when the institution did not fully understand what it was supporting. Westerman worked at the University of Delaware, where his advisor John Elias co-invented the multitouch system and co-founded FingerWorks. Faggin worked at Fairchild, then Intel, then Zilog — each institution providing the fabrication facilities, the engineering teams, and the commercial contracts that made his designs manufacturable. Zaharia and Armbrust worked at UC Berkeley's AMPLab, an academic research lab funded by a consortium of industry sponsors. Brandenburg worked at Fraunhofer. Wilson worked at Acorn. Herman worked within the aerospace and satellite industry's institutional infrastructure. The Palantir builders worked first within the intelligence community, then built a company with the explicit support of In-Q-Tel, the CIA's venture capital arm.

None of these were solo acts. The myth of the lone genius working in a garage is one of the most persistent and least accurate narratives in the history of technology. Garages existed — Hewlett-Packard, Apple, Amazon — but what made those garages productive was not isolation. It was proximity to institutions: Stanford, Xerox PARC, the logistics infrastructure of the Pacific Northwest. The inventors in this series were embedded in institutional contexts that provided funding, facilities, colleagues, and — crucially — problems worth solving. Straubel's battery work was enabled by Stanford's solar car programme. Zaharia's Spark was enabled by Berkeley's tradition of systems research. Brandenburg's MP3 was enabled by Fraunhofer's long-term research funding model, which allowed him to spend years on a problem that had no commercial customer until the internet made it one.

The institutional context matters because it determines which problems get worked on, which solutions get built, and which inventors get the decades of runway they need to compound their expertise. Shrink the funding, shorten the timelines, demand quarterly deliverables, and you select against the kind of work that produced the microprocessor, the ARM instruction set, the MP3, and Spark. These were not eighteen-month projects. They were careers. The institutions that supported them understood — sometimes accidentally, sometimes deliberately — that foundational technology requires patience measured in decades, not quarters.

A Note from the Author

We are VastBlue Innovations, a company that builds agentic AI systems for core industries — energy, utilities, manufacturing, financial services — from our office in Funchal, Madeira.

We wrote this series because we believe the best way to understand where technology is going is to understand where it came from and who built it. Not the press releases. Not the keynotes. Not the funding announcements. The actual engineering — the theses, the patents, the architecture decisions, the problems that hurt enough to demand solutions.

The editorial methodology behind The Inventors reflects VastBlue's broader analytical approach. We read patents. We read doctoral theses. We read the original technical papers, not the summaries of summaries that constitute most technology journalism. When we profiled Westerman, we read his 364-page thesis. When we profiled Faggin, we studied the MCS-4 technical documentation and the silicon gate patent filings. When we profiled Brandenburg, we worked through the psychoacoustic models described in the ISO MPEG-1 Audio Layer III specification. This is not a boast — it is a statement of method. We believe that the primary sources contain information that secondary sources systematically lose, and that the difference between reading a patent and reading a blog post about a patent is the difference between understanding a technology and having an opinion about it.

This matters because VastBlue's commercial work depends on the same kind of deep technical understanding. When we build agentic AI systems for energy companies managing grid infrastructure, or for manufacturers optimising production workflows, we need to understand the underlying systems architecture — not at the level of a product demo but at the level of data models, control loops, and failure modes. The editorial programme and the engineering practice share a conviction: that depth of understanding is not optional, and that the effort required to achieve it is the barrier that separates consequential work from noise.

The best way to understand where technology is going is to understand where it came from, and who built it. Not the press releases. The patents. Not the keynotes. The theses.

VastBlue Editorial

The people profiled here are not abstractions. They are engineers who sat at desks, wrote code, drew circuit layouts on light tables, filed patents, founded companies, and — in every case — stayed with their problem long enough to solve it. The common thread is not brilliance, though several of them are brilliant. It is persistence applied to a real constraint. Westerman's sore hands. Straubel's dead batteries. Zaharia's slow disks. Brandenburg's too-large audio files. Herman's unreachable geographies. Lyras's drowning paperwork. The Palantir builders' unconnected databases. The constraints were specific. The solutions turned out to be universal.

At VastBlue, we study patents, analyse architectures, and build systems that connect industrial data — because we believe the next generation of consequential technology will come not from consumer apps but from making the physical systems that society depends on work better. Energy grids that balance renewable intermittency in real time. Manufacturing lines that detect quality deviations before they become defects. Utility networks that predict infrastructure failures before they cause outages. These are not glamorous problems. They are important ones. And they are the kind of problems that reward exactly the approach the inventors in this series exemplify: specific constraints, deep domain knowledge, compounding expertise, and the patience to stay with a problem until it yields.

The editorial programme you are reading is part of that conviction. If it has made you think differently about one person, one patent, or one architectural decision, it has done its job. If it has made you suspicious of innovation narratives that skip over the engineering — that talk about disruption without explaining the circuit, that celebrate the founder without crediting the thesis — then it has done more than its job.

What Comes Next: The Operators

Series 2 — Reindustrialising Europe — will examine how Europe rebuilds its industrial base in the age of AI. If The Inventors was about who built the foundations, Reindustrialising Europe is about who builds on them next — and where.

The series will profile what we call "The Operators" — the people and institutions turning European invention into European industry. We will examine the structural factors that have historically separated European research excellence from European commercial scale: the capital gap, the single-market fragmentation, the risk-appetite asymmetry between publicly funded research and venture-backed commercialisation. We will profile the companies, policy frameworks, and infrastructure investments that are attempting to close those gaps — from the European Chips Act to the green industrial policy reshaping energy infrastructure across the continent.

The thesis of the second series is that invention without operation is incomplete. The inventors profiled in this first series built foundations. But foundations need buildings. The microprocessor needed the personal computer industry. The ARM instruction set needed the smartphone ecosystem. The MP3 needed the digital music distribution infrastructure. Spark needed the cloud computing platforms that made distributed computation accessible. In every case, the distance between the foundational invention and the industry it enabled was bridged by operators — people who understood the technology deeply enough to deploy it at scale, in real markets, against real constraints.

Europe's challenge in the coming decade is not invention. It never was. The challenge is operation — building the companies, supply chains, capital structures, and regulatory frameworks that turn European research into European industry. It is a different challenge from the one the inventors in this series faced. Equally important. Arguably more urgent. And it is the subject of our next eleven episodes.

Sources

1. Episodes 1-10 of The Inventors series, VastBlue Editorial — https://www.vastblueinnovations.com/editorial/the-inventors
2. VastBlue Innovations company information — https://www.vastblueinnovations.com/about