DeepSeek Déjà Vu: Why We Keep Forgetting What We Already Know
The market's predictable panic over DeepSeek's AI model release highlights our cyclical amnesia about technological progress. Each wave of cheaper, better AI sparks the same reactions we've seen before, while missing the bigger picture of inevitable commoditization. As AI capabilities become utilities, the real disruption still lies ahead.
Photo Credit: Rob GrzywinskiOriginally posted on January 28, 2025 on LinkedIn. Edited from original version.
We've Seen This Movie Before: The Alpaca Story
Remember March 2023? Stanford released Alpaca, and suddenly everyone realized you could build a pretty decent AI model for less than the price of a decent gaming PC. The tech world gasped, stock prices wobbled, and then… we all sort of forgot about it.Fast forward to January 2025. DeepSeek drops their model, performs neck-and-neck with OpenAI, and – surprised Pikachu face – NVIDIA's stock takes a nosedive. Dogs and cats living together. Mass hysteria.
Market Panic: When AI Gets Cheaper (Again)
Here's the thing about collective amnesia in tech: we're like goldfish swimming in a bowl of Moore's Law, constantly surprised by our own reflection.Let's break down this recurring fever dream:
Someone proves AI can be done cheaper/better/faster
Markets panic
Tech giants wobble
Everyone forgets
Repeat
But here's what's really happening: each of these "shocks" is just another crack in the dam. DeepSeek isn't important because it's DeepSeek – it's important because it's another sign that AI capabilities are becoming commoditized faster than you can say "market capitalization."
The Commoditization Countdown: Why Every AI Company is a Future Utility
The real story isn't about benchmark scores or stock prices. It's about how every time this happens, we're getting closer to the moment when AI capabilities become as ubiquitous as electricity. You don't buy "premium electricity" from a luxury power company. You just plug things in.And while everyone's obsessing over whether DeepSeek is the "OpenAI killer" (it's not), they're missing the bigger picture: these models are becoming utilities. The future isn't about who has the best model – it's about who owns the endpoints where these capabilities get deployed.So here we are, watching the market throw another tantrum about something we literally saw coming two years ago with Alpaca, five years ago with GPT-3, ten years ago with deep learning…
The Real Disruption Hasn't Even Started
The punchline? This isn't even the final act. It's just another rehearsal for the real show: the complete commoditization of AI capabilities. And when that happens, the real question won't be who has the best model – it'll be what we do when everyone has access to all of them.Welcome to the future. Again. Try to remember it this time. P.S. - And to all those middleware companies currently pivoting to become "AI-first": I hate to break it to you, but you might want to start learning about utility companies instead.
The development of artificial intelligence mirrors the Manhattan Project's urgency and uncertainty, but unfolds in full public view. This unprecedented transparency creates a paradox where more information leads to less clarity, forcing us to confront deep questions about intelligence, consciousness, and our comfort with the unknown.29 January 2025
The path from unfocused dialogue to meaningful collaboration requires carefully crafted personas that shape how LLMs think and respond. By defining clear boundaries and expertise domains, we create AI partners that offer consistently valuable insights rather than merely impressive but generic responses.27 January 2025
