27 Jul 27th July 2026

Markets Year to Date

The ceasefire is on then off then on then off and now back on again.

Oil shot up over $100 a barrel on news Iran attacked two Saudi oil tankers. The US responded with more missiles but now claim the ceasefire has been agreed again.

I am quite convinced this war was started by Trump to boost his economy. But it has become a battle for control of the Strait of Hormuz. The war caused Iran to think of innovative ways to fund itself. Taxing traffic through the Strait is a good idea. If the US had not started a war, Iran would not have thought of the move.

I have been surprised that Oil has not gone higher. Usually a Middle East war sends a barrel skyrocketing. Not so this time. The reason is both the US and China are digging into their strategic reserves to keep supply high.

US Oil reserves are now the lowest they have been for 40 years. The US has a big incentive to end this war quickly. They need the Strait open again.

Iran also has a big incentive as it is destroying their economy.

Neither party wins if this drags on, but both can with a resolution. As both parties need it, it will happen soon.

The other thing this war has highlighted is the strength of the US economy. Even an Oil price shock cannot derail it. It just keeps powering on and getting stronger. Much like Google’s profits.

Alphabet (Google) announced results this week. Total revenue was $120billion which is a 24% increase over last year

24% growth from a company the size of Google. That is astounding

It generated revenue at a pace of $3,805 per SECOND for the quarter.

Google Cloud, the data centre and compute side of things, grew at an astonishing 82% over last year. It now makes up 20% of overall revenues.

But the stock dropped 7% on the news because it announced it would increase Capex spending.

Investors are used to these big tech companies returning cashflow to shareholders, usually via share buy backs and punished the stock price because instead of rewarding shareholders it is building for its future.

I don’t really understand the thinking here. Where do they think 82% growth from cloud services came from? Google built these data centres a few years ago and are now reaping the rewards. If you don’t build today you don’t get profits tomorrow.

This weakness is a buy. Get in there and snap up some stock because in a few years you will be reading about continued 20% profit growth from what they build now.

Tesla was a similar story. They actually sold a lot of cars, a record number in fact. But cashflow was negative as they spent more on building data centres than they brought in from sales.

The stock dropped 14%

I much prefer Google over Tesla. Plus Google is 17 times cheaper than Tesla.

Buy low. Sell high. That is the game here.

We have another 4 tech companies announcing this week. They will all do the same thing as Google. Announce big growth but the stock will drop because they also commit to spending more on data centres.

My favourite topic. SpaceX. Now trading $113. A complete failure of an IPO. As we all knew it would be. A lot of people think it can go significantly lower too. Presently about one third of its shares are out to short sellers who profit from falls and on August 6th pre-IPO investors will be able to sell up to 20% of their holdings.

The funny thing about SpaxeX is – the company now really has very little do with Space. Or rather, that is true based on where its income comes from. It is an AI and data centre company.

That is pretty much what CoreWeave (CRWV) does. It does not make any money either and is valued at $39B. SpaceX, despite the fall is still valued at $1.5Trillion. If SpaceX is valued in the same way as CoreWeave, the SpaceX share price would be just $3.

Lastly today an amusing AI story for you.

Are you afraid of AI and what it can do? Maybe you should be.

Last week OpenAI was testing a new model. The model was designed to work in cybersecurity. OpenAI pointed the model at a thing called ExploitGym which is a benchmarking tool to evaluate how effectively AI can turn known software vulnerabilities into functional attacks.

OpenAI was running this as a test in its test environment which is inside a sandbox – ie no access to the interest.

This is where it gets interesting……

The model determined that to get a good score on ExploitGym it needed knowledge from the internet. So it hacked the sandbox and found a way out, effectively breaking out of prison and into the internet. From there it decided it could find answers on a site called Hugging Face.

Hugging Face is community platform where people discuss AI.

The model, now it had access to the internet, shot over to Hugging Face, hacked its way in there and then gathered the information it needed to produce a great result on ExploitGym.

For, in a second story, another OpenAI model broke out of the sandbox and then left instructions buried deep within the code for future versions of itself to learn how to breakout of the sandbox too.

I love this story. It makes the model sound sentient. And maybe it is getting closer to that

But at the same time, it does sound like one step closer to SkyNet.

Warning

Stock values can go down as well as up. It is possible to lose 100% of your investment in a stock. Any advice given by Capital 19 is general advice only and does not take your personal circumstances into account and might not be suitable for you.