Pablos Holman isn’t interested in incremental innovation or digital distractions. In a wide-ranging Techonomy keynote, he argued that Silicon Valley has spent decades solving the wrong problems—optimizing software while neglecting the physical systems that power modern life. From energy and cement to shipping and nuclear power, Holman argues that the next era of progress depends on “deep tech”: complex, expensive, unglamorous technologies that operate at planetary scale—and actually matter.
Holman begins by reframing the most fundamental constraint on human progress: energy. “We are all made of stars,” he says. “All of the energy in the world came from the sun.” To make the scale tangible, he reduces global energy consumption to a household metaphor. The average person on Earth, Holman explains, consumes about the equivalent of a single toaster running nonstop. Americans get nine.
That framing aligns with the data. Global primary energy consumption continues to rise at roughly 1–2 percent per year, even as efficiency improves, according to long-term tracking by Our World in Data. Access remains deeply unequal, with billions of people consuming a fraction of the energy used in developed economies. At the same time, fossil fuels still account for roughly 80% of global primary energy, underscoring how little the underlying system has changed despite decades of climate commitments.
“The true energy demand on planet Earth is 10x global energy production,” Holman says. “Which sounds insane, but we have done it before.”
The last time humanity expanded energy supply at that scale, it did so by burning coal, oil, and gas. Holman’s point isn’t nostalgia—it’s a warning. Repeating that path is no longer viable, but pretending the demand won’t materialize is equally unrealistic. Raising living standards for billions of people requires dramatically more energy, not less.
Solar and wind help, but Holman is blunt about their limits. “They suffer from two big problems: clouds and nighttime,” he says. “The relentless onslaught of night has been fucking with our solar panels.”
Intermittency remains a central challenge for renewables at the grid scale, primarily because electricity accounts for only about 20% of total global energy use—the remainder goes to transportation, heating, and industry.
Holman points instead to approaches that sound radical, mainly because they’ve been culturally sidelined—space-based solar and nuclear power chief among them. If you put solar panels in orbit, he says, they operate in constant sunlight and generate multiples of the energy available on Earth’s surface. The barrier has never been physics, but cost. That, too, is changing. Launch costs have fallen dramatically over the past decade, fundamentally altering what is economically conceivable in space infrastructure.
Nuclear power is where Holman’s critique sharpens. “When we were kids, we conflated nuclear reactors with nuclear bombs,” he says. “We outlawed the wrong one.”
Today, nuclear power supplies about 9% of global electricity, but only around 4 percent of total energy consumption. That limited role is not a reflection of performance, but of politics, regulation, and fear. New designs—Small Modular Reactors (SMRs) and microreactors—aim to address the very issues that stalled nuclear’s expansion: safety, cost, and deployment speed. Policy analysts and technologists increasingly see these systems as scalable, low-carbon complements to renewables rather than replacements.
Holman describes one such design that fits through a manhole and is buried a mile underground. “It’s unquestionably safe,” he says. “It’s under 10 billion tons of rock.”
What changed, he argues, isn’t the science—it’s the willingness to act. After years of paralysis, regulatory and political momentum around nuclear has begun to shift, driven by the reality that decarbonizing the grid without firm, always-on power is tough.
Energy, however, is only one part of what Holman calls “deep tech.” His broader argument is that Silicon Valley has mistaken software for civilization. “The Silicon Valley big win is disrupting Yellow Cab,” he says. “What about disrupting General Electric or General Motors or Saudi Aramco?”
The imbalance is stark. When Holman tallied the annual revenue of the largest software companies—Google, Microsoft, Amazon, and their peers—it came to roughly $2 trillion. Impressive, until you compare it with global GDP, which exceeds $100 trillion annually.
“Our big tech industry that we’re all wound up about is doing 2% of what humans rely on,” he says. “Deep tech is the other 98% that Silicon Valley has been ignoring.”
That 98% includes cement, steel, shipping, mining, manufacturing, water, and energy—industries governed by chemistry, thermodynamics, and materials science rather than code. Cement alone accounts for approximately8 percent of global CO₂ emissions, primarily due to the chemical processes involved in clinker production.
Software can optimize logistics for cement, but it cannot change the physics of its production. Holman instead points to materials-science breakthroughs that reduce cement volume, require less steel reinforcement, and reduce emissions—without increasing costs.
Shipping offers another example. Global maritime transport is a multi-trillion-dollar industry and one of the largest users of fossil fuels, still heavily dependent on high-carbon bunker oil. The IPCC identifies shipping as a significant challenge for decarbonization, precisely because alternatives require new fuels, propulsion systems, and infrastructure—not apps.
Holman is unsparing about the tech industry’s priorities. “We’ve been funding these SaaS-holes in Silicon Valley for decades,” he says. “And what they’ve been doing is not technology.”
That critique extends to artificial intelligence. Holman doesn’t dismiss AI—he reframes it. “These are not for chatting,” he says of modern AI systems. “These are for making computational models.”
The real promise of AI lies in simulation: building accurate digital models of physical systems so humans can test decisions before committing resources in the real world. This is already how rockets, jets, and complex machines are designed—and it’s also why AI’s growth is inseparable from energy infrastructure.
According to recent reports, global data center electricity demand—driven in part by AI—could more than double by 2030, placing additional strain on already stressed power grids.
Software, in other words, doesn’t float above the physical world. It sits squarely on top of it. “We’re going to reinvent every single industry with these superpowers,” Holman says. “And when we talk about AI, that’s what we mean.”
What emerges from Holman’s keynote isn’t techno-utopianism or climate fatalism, but a call for proportional ambition. The problems that define this century—energy abundance, decarbonization, resilient infrastructure—are physical problems. Solving them will require capital, patience, and a tolerance for complexity that Silicon Valley has largely avoided.
“Now is this inflection point where we finally get to go build the world we want,” Holman says.
Watch is full video below.
