Stick with me here…

This is computing in 1979 in a nutshell…

Using BASIC code:

  • 10 HGR
  • 20 HCOLOR = 3
  • 30 FOR A = 0 TO 6.28 STEP 0.05
  • 40 X = 140 + 40 * COS(A)
  • 50 Y = 96 + 40 * SIN(A)
  • 60 HPLOT X, Y
  • 70 NEXT A

And here’s what each of those command lines did:

  • HGR switches the Apple II into its 280 × 192 high-resolution graphics mode.
  • HCOLOR = 3 selects white.
  • A runs from 0 to roughly 2π2\pi, going all the way around a circle.
  • COS(A) controls the horizontal position.
  • SIN(A) controls the vertical position.
  • (140,96) is almost exactly the centre of the screen.
  • The 40 gives the circle a radius of 40 pixels, so it’s about 80 pixels across.
  • HPLOT X,Y plots each calculated point.

And if you got all that perfectly correct, here’s what you’d see:

What would you see?

That’s it.

That’s the whole 1979 experience.

Seems so… basic now. But in 1979 (for those using computers then) it would have been magical.

Nearly five decades later, and we’re a heck of a lot further down the computing road.

Now, anything you can think of, you can ask a computer to do.

A little more specifically, you can ask AI to do.

In fact, as of last week, we may now be at a point where there isn’t anything AI can’t do.

That’s quite a claim… but if we are where Jensen Huang (Nvidia CEO) thinks we are, then humanity is about to enter an era where there are no limits to what we can achieve.

“AGI has arrived”

Last week, OpenAI released its latest model, “Astra.” It came to us with a snazzy, full-of-retro-vibes ad.

OpenAI post

I do recommend checking it out here.

OpenAI calls it their most intelligent model yet, state of the art at using a computer, writing software, science, and cybersecurity.

It’s also the first model to hit “Critical” on OpenAI’s own cyber scale, meaning it can find unknown security flaws in well-protected systems without a human guiding each step.

And their ad takes a very distinct look at how far computing has come in since 1979.

Then on Sunday, Jensen replied to the news on X…

Birthplace of AGI?

“AGI has arrived.”

So, what is AGI?

Artificial general intelligence is seen as somewhat of an end point, and a starting point, for AI.

The end of trained human-input AI, and the start of it cracking on and doing its own thing.

Albeit, OpenAI’s charter defines it as highly autonomous systems that outperform humans at most economically valuable work.

It’s a system that can do your job… and mine.

Of course, Jensen sells the GPUs, so in calling it AGI, he’s also saying, buy more of my GPUs and AI chips!

And that’s also what the market heard. 400,000 GPUs coming online. That’ll just be the start of it. And if that’s the start of it, then the rest of the AI hardware market, from the big end all the way down to the microcap component makers, is set for another massive AI market melt-up.

From answering to doing

I also think it’s worth noting that Astra comes hot on the heels of xAI releasing Grok Bot.

Grok Bot gives you always-on AI “teammates,” and each one gets a cloud computer of its own, logs into your actual software, and carries jobs through end to end while you sleep.

When the internet took off in the 1990s, which piece of infrastructure experienced one of the most spectacular investment booms?


Find out what your friends and family think as well. Share this email with them so they can answer today’s poll and subscribe as well.

Hot US jobs data knocked the S&P 500 down 0.4%, and yet the AI hardware names ignored it completely. SanDisk (Nasdaq: SNDK) ripped 11.9% higher, SK Hynix rose 8.1%, and Micron (Nasdaq: MU) gained 6.1%, breaking back over US$1,000 per share.

The fear all year has been that the AI trade is overstretched, a house of cards built on funny money, and that the capital expenditures will never pay off.

I think that’s a backwards view. The rollout isn’t under threat, it’s accelerating…

Merely months ago, Anthropic jumped from Opus to Fable, and there’s still Mythos that’s being held behind the gate. OpenAI has gone from GPT-5.3 to 5.4, 5.5, 5.6, and now into GPT-6 Astra, “A new generation of intelligence,” and all that has happened just in 2026.

This acceleration is very real, and the companies supplying compute, memory, storage, and power are staring at years of demand they can’t yet meet. Every major tech player is getting into this and extracting full value from it, be it Google, Meta, Microsoft, Amazon, and, as we’ll see in just a few days’ time, Apple as well.

The giants are already priced for a lot of that. But the smaller companies feeding the same machine aren’t…

And that’s where the asymmetric opportunities sit. It’s exactly the corner of the market my research targets, and if you want the small caps I think are best placed for this acceleration, start here.

Until next time,


Sam Volkering
Investment Director, Southbank Investment Research