# VC10X Pulse - AI v/s Humans - The debate we aren't ready for

An episode of [VC10X - Investing, Venture Capital, Asset Management, Family Office, Wealth Management](https://invisiblenetwork.ai/podcast/1632806986), 26 Feb 2026, 7:21.

Audio: https://anchor.fm/s/a453ddd4/podcast/play/116078831/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-1-26%2F418868813-44100-2-7e58739fd3f2a.mp3

Sam Altman recently defended AI's massive energy consumption by saying humans are energy-intensive too — consuming energy for 20 years before becoming useful. We think it deserves a harder look. In this episode, we break down why that comparison is more than a clever deflection. We examine what AI's learning actually runs on and why that changes the argument entirely, why reducing humans to an efficiency metric is a framing worth resisting, and why this particular narrative is one that venture capitalists find very easy to fund — which tells you something important in itself. We also look at what this means practically for investors: how to stress-test civilisational storytelling, which regulatory risks are being underpriced, and why energy intensity is a structural cost question as much as an environmental one. No drama. Just clear analysis of a soundbite that slipped past most people — and probably shouldn't have. LINKS Prashant Choubey - https://www.linkedin.com/in/choubeysahab Subscribe to VC10X newsletter - https://vc10x.beehiiv.com Subscribe on YouTube - https://youtube.com/@VC10X Subscribe on Apple Podcasts - https://podcasts.apple.com/us/podcast/vc10x-investing-venture-capital-asset-management-private/id1632806986 Subscribe on Spotify - https://open.spotify.com/show/7F7KEhXNhTx1bKTBFgzv3k?si=WgQ4ozMiQJ-6nowj6wBgqQ VC10X website - https://vc10x.com For sponsorship queries reach out to prashantchoubey3@gmail.com SUBSCRIBE FOR MORE VC10X breaks down the most important stories in finance, tech, and markets every week. Subscribe for actionable insights.

## Transcript

[0:00] There's a certain kind of dangerous idea that doesn't
[0:02] announce itself as dangerous. It arrives dressed as weight, It
[0:06] gets a laugh, people move on, and the idea quietly,
[0:10] efficiently gets into the water supply.
[0:13] Sam Ortman had one of those moments recently.
[0:15] Asked about AI staggering energy consumption, he said humans are
[0:19] energy intensive too. They consume energy for 20 years
[0:22] before they become useful. The room probably nodded in
[0:26] agreement, and most people moved on.
[0:29] But there's something serious there.
[0:31] Because in that one sentence, Altman did three things
[0:35] simultaneously. He reframed humans as an
[0:37] efficiency problem. He positioned AI as the solution
[0:40] to that problem, and he built a valuation story that venture
[0:44] capitalists would find very, very easy to find.
[0:47] Let's take that apart. Let's start with the argument on
[0:51] its face, because it deserves a fair hearing before we challenge
[0:55] it. Odd man is drawing a cost
[0:57] benefit analogy. AI consumes energy, yes, but so
[1:01] do humans during childhood and education.
[1:04] Therefore the energy cost of AI is justified.
[1:07] And to be fair, AI does learn while it consumes that energy,
[1:11] just like humans. It trains, it iterates, it gets
[1:15] more capable. That part is real.
[1:17] But here's what that learning actually runs on, and this is
[1:21] the part that gets quietly omitted.
[1:24] AI's learning runs entirely on what humans already produced
[1:28] during their 20 years or their entire history.
[1:31] The literature, the science and the legal reasoning, the medical
[1:35] knowledge, the philosophy, the art, every data set that makes a
[1:39] model capable is the accumulated output of human development at
[1:43] civilization scale across centuries.
[1:46] AI doesn't learn instead of that process, it learns because that
[1:49] process happened billions of times over, long before the
[1:53] first model was trained. So the analogy has a hidden
[1:56] dependency, but it inside. You cannot compare a human child
[2:00] with an AI system as equivalent energy inputs when one of them
[2:04] is only capable of learning because countless millions of
[2:07] other kind lived full lives and left something behind.
[2:12] And there's the second problem here, perhaps a more important
[2:14] one. A child's 20 years doesn't just
[2:17] produce capability, it produces moral agent, someone who will be
[2:22] accountable for their decisions, someone who belongs to a
[2:25] community, bears consequences, and can be held accountable.
[2:29] The energy going in is not overhead waiting to become
[2:33] productive. It is a formation of a person
[2:35] who will carry the weight of the world.
[2:37] What AI produces, however sophisticated, is capability.
[2:41] Without that accountability at massive scale.
[2:43] Comparing the two as equivalent inputs ignores the most
[2:47] important output entirely, that moral compass.
[2:51] But let's go one level further, because the comparison doesn't
[2:54] just misinterpret a is learning. It does something more corrosive
[2:58] to the broader conversation. When you frame a human being as
[3:01] an energy input that takes 20 years to generate a useful
[3:04] output. You have reduced a person to a
[3:07] return on an investment calculation.
[3:09] The follow on questions become logical rather than monstrous.
[3:13] What if the ROI is poor? What if a person never becomes
[3:17] useful by that metric? What if AI simply does it
[3:20] faster, cheaper, and without the 20 years of overhead?
[3:24] This is not a hypothetical philosophical concern.
[3:27] This is an active narrative being constructed that AI is not
[3:31] a tool for humans, but a replacement model for humans.
[3:34] At least that is something that you can or people are making of
[3:37] that statement and the language Oddman used, whether
[3:40] deliberately or carelessly, advances that narrative.
[3:44] Innovation has always had a contract with humanity.
[3:47] The question has never just been can we build this?
[3:50] It has always been who does this serve and at what cost to whom?
[3:54] Nuclear energy is powerful. Social media is engaging.
[3:58] Efficiency is not the same as good, and a technology that is
[4:01] framed as superior to the species it supposedly serves
[4:04] should at minimum be asked to justify that framing rather than
[4:08] have it sail past the moment we accept a framework where humans
[4:11] and AI are compared on energy efficiency, we have already
[4:14] conceded something that should have never been on the table,
[4:18] the assumption that technology exists for people, not instead
[4:22] of them. Now let's ask the question we
[4:24] always ask Who benefits from this framing?
[4:28] Open AI is seeking capital, enormous amounts of it.
[4:31] The energy infrastructure alone requires investment at a scale
[4:34] that rivals major industrial projects.
[4:36] To attract that capital, you don't just need a product story,
[4:39] you need a civilization story, the kind that makes an investor
[4:42] feel they are not funding a company but participating in the
[4:44] next stage of intelligence itself.
[4:47] Comparing AI to human development is exactly that
[4:50] story. It says this is not a technology
[4:52] project, this is a more efficient version of humanity.
[4:56] And if you accept that, if you let that framing in, then the
[4:59] valuation becomes easier to justify, the energy costs become
[5:03] easier to absorb, and the ethical questions become much
[5:06] easier to differ. This is not a conspiracy.
[5:09] It is a rational behavior for someone in an old man's
[5:12] position. He's a builder with a mission, a
[5:14] company to fund, and a narrative to protect.
[5:16] But our job as allocators, as investors, as people responsible
[5:20] for other people's capital, is to see the architecture behind
[5:24] the sound bite. A compelling story and a sound
[5:26] investment thesis are not the same thing.
[5:28] And a story that venture capitalist love is not
[5:31] automatically one that the society should accept
[5:34] uncritically. O what do we actually do with
[5:37] this? Three things.
[5:38] First, hold the frame when AI comanies present their value in
[5:42] civilizational terms. Return to the Simler questions.
[5:45] What does the unit economics look like?
[5:48] What is the credible part to sustainable energy use?
[5:51] Who absorbs the cost of getting there?
[5:53] The company, the grid, or the public?
[5:55] Narrative is not a substitute for fundamentals, no matter how
[5:59] well delivered. Second, what's the regulatory
[6:02] environment? Seriously, the comparison Ottman
[6:04] made is precisely the kind of argument that will be tested in
[6:08] legislative chambers and international forums over the
[6:10] coming decade. Jurisdiction that push back on
[6:13] the streaming, and several already are, will create
[6:16] meaningful headwinds for companies that have priced in
[6:19] frictionless expansion. That is a risk worth modelling,
[6:22] not dismissing. The energy intensity of AI is
[6:25] not only an environmental question, it is a structural
[6:29] cost question. Any business model where input
[6:31] costs are this large and this concentrated carries real
[6:35] vulnerability. Compute efficiency
[6:37] breakthroughs, green energy dependencies, regulatory shifts
[6:40] on data use, all of these affect the underlying economics in ways
[6:44] that a confident answer at a conference does not insulate
[6:47] against. Sam Altman is a serious builder.
[6:51] That is not in dispute. But serious technology operated
[6:54] at this scale, with this much capital and this much influence
[6:57] over how we talk about human beings deserves serious scrutiny
[7:01] in return. When someone compares a child to
[7:04] an energy Ledger and gets an applause for it, the right
[7:07] response is not to move on. It's to ask what kind of future
[7:11] is being built, by whom, and on who's terms.
[7:14] That's it for this week. We'll be back next week with
[7:16] another story from the front lines of global investing VC10X.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.