👇 AI scales down

What’s going on here?

AI companies and tech titans have been banking on a new way to drive big revenue: small language models.

What does this mean?

Tech giants Apple, Microsoft, Meta, and Google have been releasing “small language models” lately, eager to woo businesses that are wary of the massive costs and power demands of beasts like OpenAI’s ChatGPT. See, GPT-4 and Google’s Gemini 1.5 Pro are both large language models, the smartest type on the market. Problem is, due to copyright and data issues, they’re expensive. The smaller versions, meanwhile, are a winning combination of cheaper, more customizable, and energy-efficient because they require less power to train and run. Not to mention, they can process tasks locally on a device – perfect for those who want to keep data in-house and not in the cloud.

Why should I care?

Zooming out: The ins and outs.

Currently, when you ask models like ChatGPT to pick an outfit or write a poem, the request zips off to OpenAI’s servers in the cloud, gets processed, and bounces back with an answer. That requires internet access and shares data with the model maker. So “edge computing”, where everything happens right on your device, is creating a buzz – and driving demand for speedy “edge” chips that can run models without spilling your data. The chips need to be small, cheap, and energy-efficient so they can fit into a device like a smartphone. Enter Apple, which is scrambling to develop its own AI chip to upgrade the iPhone.

The bigger picture: Small but mighty.

If edge chips get small and cheap enough, AI-driven smart gadgets will be everywhere – from homes to offices. Nowadays, cloud-based AI models are mostly confined to data centers, which demand a ton of power and space. So if AI could run directly from folks’ devices, those challenges may become a thing of the past.

What’s going on here?

Gold set a new record high on Monday, breezing past the benchmark it set in April.

What does this mean?

Gold prices have been ticking up since October 2022, but they really took off this March. None of the usual culprits seem to be behind the scale and velocity of the rally: the economy has been strong, interest rates are high, and the US dollar is holding firm – all of which would usually make gold’s price more modest. But right now, global political tensions are at a boiling point and US debt is disconcertingly high, pushing investors toward the “safe haven” of gold. After all, the precious metal maintains its value even when currencies don’t, can hedge against bonds and stocks falling, and doesn’t default. It will also have helped that central banks have tripled their gold purchases since Ukraine was invaded two years ago.

Why should I care?

For markets: Silver wins gold.

Gold has grabbed the headlines this year, but silver is the one racing ahead. The two precious metals are similar: they’re both safe-haven assets that can hedge against inflation. So it’s no wonder silver has followed its cousin higher. But silver is also seeing strong industrial demand in clean energy, since it’s particularly handy for infrastructure like solar panels. Plus, it currently takes about 80 ounces of silver to buy one ounce of gold, higher than the 20-year average of 68. So in relative terms, silver is still cheap.

The bigger picture: Slipping and sliding.

Many investors are worried that high levels of government debt across the globe are unsustainable. The fear is that policymakers might have to spur on inflation to decrease the debt’s value over time, which could push countries’ currencies down. Mind you, with gold prices climbing in almost all currencies, there’s reason to believe that they’re already losing value.