Get AI summaries of any video or article — Sign up free
The unhinged world of tech in 2026... thumbnail

The unhinged world of tech in 2026...

Fireship·
5 min read

Based on Fireship's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.

TL;DR

LLMs are portrayed as plateauing after GPT-5, but AI momentum continues through robots, wearables, and the compute systems that support large-scale deployment.

Briefing

2026’s biggest tech story is a shift from “AI that writes code” to “AI that runs the world”—with robots, wearables, and massive compute demand driving the next wave of investment, jobs, and regulation. LLMs may have plateaued after the GPT-5 letdown, but the broader AI economy is still expanding through automation of physical work, new interfaces, and the infrastructure needed to keep models fed. That combination matters because it changes where money flows (chips, power, data centers) and what kinds of work remain (human oversight and cleanup rather than pure creation).

Software employment is unlikely to snap back to 2023 levels, even as the Bureau of Labor Statistics forecasts about 15% growth for software development jobs through 2034. A policy change to the H-1B program adds a $100,000 fee for applicants, making it harder for U.S. companies to hire lower-cost overseas talent. Meanwhile, AI coding tools are still far from replacing engineers; instead, they’re generating “code janitors”—developers tasked with cleaning up the slop, bugs, and maintenance debt produced by automated coding.

The AI hype cycle is also entering a new phase. LLMs have stopped delivering exponential leaps in intelligence, and disappointment around GPT-5 signals that “smarter-than-human” progress isn’t arriving on schedule. Still, the bubble may not be near its end because many AI companies remain private and are positioned for a future IPO wave. SpaceX, OpenAI, and Anthropic are cited as potential candidates for major public listings in 2026—an event that would likely mark the transition from VC-backed storytelling to public-market scrutiny.

Robots and compute-heavy infrastructure are expected to keep momentum alive. Humanoid robots like 1X’s Neo (laundry and dishwashing) and factory automation efforts tied to Figure Robots and Tesla Optimus are framed as the next consumer and industrial push, even if the technology still requires refinement. Wearables are another battleground: Rabbit and Humane Pin are labeled flops, while OpenAI’s collaboration with Johnny IV and Nike’s battery-powered shoes are treated as early bets on practical, mainstream adoption.

Power and chips may become the real bottleneck. Cloud providers such as Azure struggle to secure enough electricity for Nvidia GPUs, raising the odds of renewed nuclear investment. The transcript points to China’s reactor buildout and to small modular reactor efforts like Ollo, including a deal involving Zuckerberg to place a reactor in Ohio. If data centers can get neighborhood-scale power, AI expansion could accelerate.

Finally, 2026’s “future” includes tech-adjacent governance and developer tooling. Digital IDs and central bank digital currencies are portrayed as an advancing bureaucratic push, with the European Central Bank’s digital euro pilot expected to move into its next phase by mid-2027 and full issuance by 2029. On the engineering side, Node.js improvements (TypeScript support) and bundling/runtime shifts (Dino’s module bundler; Bunjs with Postgress and Redis support) compete with ReactJS’s continued dominance. Quantum computing is also positioned as a potential disruptor: Google’s Willow chip and quantum echoes algorithm are cited as verifiable progress that could make AI’s hype look minor once quantum reaches “version 1.0.”

Cornell Notes

The forecast for 2026 centers on AI’s next phase: LLMs may have plateaued, but automation is expanding through robots, wearables, and the compute and power systems that keep AI running. Job growth in software is still projected through 2034, yet AI coding tools are creating “code janitors” who maintain and repair the garbage code produced by automation. The AI bubble may persist longer than expected because many companies remain private and could stage major IPOs in 2026, including OpenAI and Anthropic. Robots and data-center power—potentially boosted by small modular nuclear reactors—are framed as the practical drivers of investment. Meanwhile, digital IDs and central bank digital currencies continue moving forward, and JavaScript tooling keeps shifting toward faster runtimes and better build systems.

If LLMs are plateauing, what keeps AI momentum going in 2026?

The momentum shifts from “smarter text” to automation that touches real life: humanoid robots for household and factory tasks, wearable AI devices, and the infrastructure required to run large models. Even with the GPT-5 disappointment, the transcript argues the broader AI ecosystem keeps expanding because companies can still deploy AI in physical systems and scale compute—especially as demand for linear algebra and GPU-heavy workloads grows.

Why might software jobs still exist even as AI coding tools spread?

Software employment is forecast to grow (about 15% through 2034 by the Bureau of Labor Statistics). The transcript also claims AI coding tools aren’t close to fully replacing humans; instead, they generate maintenance debt. That creates a new job category—“code janitors”—responsible for cleaning up AI-generated garbage code, fixing bugs, and keeping systems stable.

What would signal the end of the AI hype cycle?

A wave of IPOs is presented as the key marker. With many AI companies still private, the transcript expects a public-market handoff from VCs to the public later on. It specifically flags the possibility of major IPOs in 2026 involving SpaceX, OpenAI, and Anthropic, implying increased scrutiny once those companies go public.

How do robots and wearables connect to the broader tech economy?

Robots and wearables are treated as the next consumer and industrial adoption layer, but they also depend on the same underlying compute and chip supply chain. Humanoid robots like 1X’s Neo and factory automation efforts (Figure Robots and Tesla Optimus) are positioned as replacements for manual labor. Wearables—after Rabbit and Humane Pin are called flops—are expected to improve via collaborations such as OpenAI with Johnny IV and via consumer bets like Nike’s battery-powered shoes.

Why does power supply become a central AI bottleneck?

As AI scales, electricity demand rises. The transcript claims cloud providers like Azure struggle to find enough power to run Nvidia GPUs. That pressure could revive nuclear power, including small modular reactors. It cites Ollo’s regulatory push and a deal involving Zuckerberg to place a reactor in Ohio as examples of how data centers might get more reliable, localized energy.

What developments in quantum computing could matter for AI timelines?

Quantum computing is framed as a potential disruptor once it reaches practical capability. The transcript highlights Google’s Willow chip and the quantum echoes algorithm, which it says ran a verifiable algorithm that surpassed supercomputers’ ability. The implication is that “version 1.0” quantum could make the AI hype bubble look less dramatic by enabling new capabilities beyond today’s classical approaches.

Review Questions

  1. Which factors in the transcript are presented as replacing “LLM intelligence gains” as the main drivers of 2026 tech growth?
  2. How does the transcript connect AI coding tools to the emergence of “code janitors,” and what does that imply for entry-level vs maintenance work?
  3. What role do IPO timing and power constraints play in the forecast for whether the AI bubble persists into 2026?

Key Points

  1. 1

    LLMs are portrayed as plateauing after GPT-5, but AI momentum continues through robots, wearables, and the compute systems that support large-scale deployment.

  2. 2

    Software job growth is still forecast (15% through 2034), yet AI coding tools are expected to shift work toward maintenance and cleanup rather than pure development.

  3. 3

    A $100,000 H-1B fee change is framed as making overseas hiring more expensive, potentially altering how U.S. tech companies staff teams.

  4. 4

    A future IPO wave is presented as the likely “endgame” signal for the AI hype cycle, with SpaceX, OpenAI, and Anthropic named as possible 2026 candidates.

  5. 5

    Humanoid robots (1X’s Neo; Figure Robots; Tesla Optimus) and wearable bets (OpenAI + Johnny IV; Nike’s battery-powered shoes) are positioned as the next adoption layer.

  6. 6

    Rising electricity demand for Nvidia GPUs could push data centers toward nuclear power, including small modular reactors like Ollo and deals such as a reactor in Ohio.

  7. 7

    Digital IDs and central bank digital currencies are described as advancing despite public resistance, with the digital euro pilot expected by mid-2027 and full issuance by 2029.

Highlights

The forecast pivots from “smarter AI” to “AI that scales in the physical world,” with robots and wearables keeping investment flowing even as LLM intelligence gains slow.
AI coding tools are expected to create “code janitors,” shifting developer value toward fixing and maintaining AI-generated code rather than writing everything from scratch.
Power constraints—especially electricity needed for Nvidia GPUs—are treated as a decisive bottleneck that could accelerate nuclear adoption via small modular reactors like Ollo.
A major IPO wave is framed as the clearest marker that the AI hype cycle is ending, with OpenAI and Anthropic specifically mentioned as potential 2026 events.
Quantum computing progress is tied to practical impact, with Google’s Willow chip and the quantum echoes algorithm cited as verifiable steps beyond supercomputer capability.

Topics

Mentioned