Energy, Not Code: Why AI's Next Bottleneck Is Physical Power
By Ray Dombrowski ·
Old business plans are obsolete; future growth depends less on software genius and more on industrial energy capacity.
The sheer pace of technological change described in the latest episode of “Invest Like the Best,” featuring Eric Vishria, suggests that everything we thought we knew about corporate strategy—and perhaps even how industries grow—is obsolete. The most striking takeaway is not a specific investment bet, but a fundamental warning: rigid adherence to outdated plans is itself destructive; as one speaker noted, "every single day that you are hitting your plan, you are destroying equity value."
Throughout the discussion, Eric Vishria systematically dismantled conventional wisdom regarding market dominance and technological scaling. He argued that common narratives—like the idea that one vendor will "eat everything" in a commodity cloud market—are often massively wrong, citing AWS’s poor reception in 2007 as evidence of this historical flaw. On AI itself, he pointed out that while the demand for intelligence is unlimited, the ultimate bottleneck is physical: energy. Furthermore, he shifted the focus from pure software to execution, noting that the modern Product Manager must now bridge customer problems with the "jagged edge" of rapidly evolving AI capabilities, a skillset far more valuable than traditional titles.
The New Rules of Economic Competition The economic landscape has fundamentally changed its competitive frontier. Where old success models relied on sticky interfaces—like databases, which were hard to migrate away from—AI is lowering those barriers, making migration "kind of trivial." This forces companies to redefine what makes a database successful, shifting the focus from mere storage capacity to criteria like zero-to-infinity scaling and transportability. Moreover, Vishria provided a sobering assessment of the physical constraints ahead: while AI requires immense infrastructure buildup (power, chips, memory), this struggle culminates in energy scarcity. This suggests that future growth is less about software genius alone and more about industrial power—the ability to reliably supply massive amounts of clean, consistent energy.
Beyond Software: The Industrial Reality The discussion naturally pivoted from cloud computing to robotics, which Vishria suggested has the potential to "dwarf what we're currently living through." This pivot is critical for anyone tracking real-world employment and industrial policy. Unlike LLMs, which are bootstrapped on internet data, AI robotics companies must instead focus intensely on gathering high-value data from physical environments—the equivalent of a specialized training curriculum rather than open-source scraping. Furthermore, the speaker stressed that achieving true market success requires far more than just technological capability; it demands genuine business model innovation and an understanding of how to sell "magic" without relying on old quota-based sales models.
The analysis is relentlessly focused on execution and physical constraints. While much of the conversation centers on venture capital philosophy—such as being a partner first, and an investor second—the underlying message for policymakers and industry leaders remains clear: success now requires unprecedented agility (building temporary "sand castles" instead of permanent "castles") and a deep understanding of real-world resource limitations.
The most significant takeaway is the forced collision between advanced digital capability and fundamental industrial reality. The hype cycle around AI often neglects energy infrastructure, supply chains, and the specialized data required to operate machines in messy, non-controlled environments. We cannot afford to treat this as solely a software problem; it is fundamentally an industrial challenge requiring massive capital allocation toward power generation and physical integration.
The next decade of economic growth will not be determined by who writes the cleverest code or captures the biggest market share on paper. It will belong to those entities—the companies, the regions, and the governments—that can solve the energy equation while effectively bridging advanced AI capabilities with the messy, complex reality of physical work.