Blain’s Morning Porridge July 15^th 2025: The AI revolution – where’s it headed?

“Sometimes the new, new, new thing becomes the wrong thing headed in the wrong direction.”

The AI revolution is exceedingly expensive. It’s become a matter of who can chuck the most amount of money at it. But that’s forgetting it’s still evolving and there is little to stop it moving in new directions.

Plus, The Morning Porridge experiments with AI generated graphs!

Be very suspicious when a bond-guy – like myself – starts to pontificate on stock market stories. From a rational perspective, the AI Industry appears to be as bubblesque as anything I’ve seen over the past 40 years! The clearest indications we’re approaching a “frothy top” include;

• The huge considerations being paid for AI start-ups (apparently on the basis often being: “we don’t know what it does, but we’ll buy it to stop someone else finding out”),
• The ginormous upfronts Meta is offering the top AI engineers and managers,
• Regulatory threats about how AI firms market their capabilities and who they are selling to,
• The number of data-centres being built,
• The look of fearful anguish on OpenAI’s Sam Altman whenever he’s asked about his relationship with Microsoft, or how the deal with SoftBank to finance his planned $300 bln plus of spending is going….
• Elon Musk’s plan to divert capital from Tesla and SpaceX into xAI – the market may think its genius, but the reality is maybe he’s cash strapped and already it’s sub-optimal in size.

It all gives the AI game that look of desperation for the next breakthrough, and uncertainty for what the next new, new thing to justify the froth might be. Meanwhile the money keeps pouring. Surely no one – except possibly Mark Zuckerberg –  seriously believes you can buy an AI monopoly on the open market?

As others have pointed out, Zuckerberg is able to chase his visions of “superintelligence” with lots of tens of billions of dollars because the other firms in the game have empowered shareholders able to demand their questions are answered! Disciplined corporate governance is less a thing when you are the controlling shareholder. On the other hand, Google seems to be quietly pulling ahead with its models and products – and much of it seems to actually work.

Potential pushback comes from three angles.

First is the growing sense that AI – at least in the USA – has become a money pit. There is a deeply embedded assumption that AI will only work out the US, it has to cost billions, that chip demand is unlimited and supply scarce, and comes with the requirement for massive energy inputs. At times the sector looks unaware of how quickly tech can evolve.

These assumptions were rocked when the Chinese hinted they can do it just as well, cheaper and with less energy when DeepSeek roiled the market in April. Now we understand China is awash with AI projects. If you want to see how that might end, take a look at the winners and losers in the EV market. (Clue: it ain’t Tesla on the podium.)

The big news this morning is that Nvidia will get its licences to sell its H20 chips into China. It may already be too late. It sounds like a last ditch effort to keep the US firms involved in the rapidly developing alternative China-AI ecosystem. Far from closing down China’s development of AI, the US chip sanctions forced it to become independent, ending US dominance. Smart money will be watching to see exactly how Chinese demand for US chips evolves.

The second issue is exhaustion with AI. Exhaustion is a common theme of every new, new thing in Big Tech. At a personal level, AI has become an opportunity for a new AI parasite class. I am being bombarded with offers to teach me how to use AI by “experts” offering their “unrivalled AI experience” in how to monetise the technology within my businesses (Windshift – the alternative assets advisory firm, and the The Morning Porridge.) Both are knowledge based (ie – me), but I freely admit both could use better organisation and logistics in terms of how I use my time, back-up and present my thinking.

That’s just one aspect of the widening AI backlash. A few months ago one of my chums told me he’d found his young son was taking advice on school from the AI on his phone, and regarded it as a “friend”. The social implications of a cohort of youngsters who trust AI better than people is truly frightening. Its why there are demands for regulation and even reversals on what AI is becoming.

The third issue is understanding what AI really is. It is not intelligence. It’s a very clever way of weighing up all the opinions out there on the global net (effectively the sum total of human experience) to come up with the most likely answer to any question. As I’ve written before – it’s much like how markets are voting machines, and do not always vote for the correct answer.

Smart money knows the markets are often wrong, but will still trade the market’s sentiment. I expect the future will feature many stories about how early AI would simply magnify bad information because of the way they work and what the learning models learnt.

The inaccuracy is based on how the models work and train. Clearly, the best AIs are the ones that best interpret the whole of human knowledge from all that data they scrape. The problem is – we’re now aware AIs are as lazy as the way human students subcontract their essays to it. AI is doing its learning not just from the whole net, but from what other AI’s have learnt! One of my AI chums, who is genuinely AI literate (I don’t pretend to be), reckons 70-80% of what ChatGPT, Grok or whatever, tells you, has probably been scraped straight off another AI!

My promise to readers of The Porridge is that it will never be written by AI. It will always remain hand-written and fresh most mornings, complete with the multiple bugs, spelling mistakes and bad grammar which are signs it remains a hand-crafted product. (All great art has imperfections.)

Blain tries to use AI

I will start using AI to illustrate my Morning thots. Trying to persuade subscribers to pay up means making it more interesting… Perhaps you can teach a very old bond dog new tricks?

I was playing around with ChatGPT yesterday and asked it to construct me a chart showing US wages vs UK wages. Exactly as expected, it confirms that (dollar adjusted) UK wages have flatlined this century while, US wages have risen by 2/3rds. UK workers earn 50% less than US workers. No Sh*t Sherlock. It’s part of my thesis on the underlying reasons for the sheer bloody misery of the UK today – which involves flogging the who country cheap to American PE and losing our financial edge.. among multiple other things.

US vs UK wages:

That’s nothing that any semi-competent analyst could not do in a short time by extracting the data from a couple of different sources, doing the FX and running it. Except that it took the AI moments. Not sure I quite trust the details, but the illustrated trend is good enough to illustrate something I knew to be true – UK workers are underpaid and therefore consume less compared to Americans

I then decided to ask it to compare US real wages (adjusted for inflation) to the value of the stock market, and productivity growth. It showed me this:

US Real Wages, Stocks and Productivity (Source: ChatGPT)

Now that really does confirm the theme I was thinking about yesterday. Real wages – thus consumption – in the US fallen. Wages have tumbled since 2020 – driving the MAGA thesis of how workers have been left behind since the sudden return of inflation following COVID and the Ukraine invasion.

Long-term productivity has advanced marginally year by year, but stock markets have surged by 350%. That means the owners of stocks are 3.5 times wealthier, but are effectively paying their poorer workers less, and have not been investing enough in growth to back up the rise in the value of the stock market.

My conclusion is simple. Wages and productivity don’t support the current valuation of the US stock market.

We know that wages, productivity and the stock market historically moved in close alignment, right up the 1990s and the beginning of the age of financialisation. They have been completely unaligned since 2008 – the global financial crisis – which spawned the distortions of QE and ultra-low interest rates from 2011, and the resulting imbalance across asset classes in the US and global economy.

My read is rising inequality, flatline wages, the QE distortions and relative asset imbalances triggered Populism – Brexit and Reform in the UK, and MAGA in the US. These are the real forces that spell crisis for Western Capitalism. (Anyone for the last few biscuits while we wait for economic Armageddon to arrive?)

The problem now in the US is how investment and the pursuit of wealth has been distorted to reflect the gains in financial assets vs the underperforming real economy. That is what Trump and MAGA really should be addressing – how to get the phenomenally wealthy component firms of the US stock market to invest in productivity in the US economy. That’s not just a matter of opening new production facilities overnight – it’s a complete shift in investment.

Trump is quite right: it should be perfectly possible for Apple to make iPhones in America. To do so today from scratch by US workers would triple their cost.

How different the world may have been If Apple, et al, had been investing in robotics, new factories and production methods for the last 30 years? Instead Apple invested profits in building its massive $200+ bln portfolio of long-term investments, plus a $49bln war-chest of short-term cash and equivalents – while maximising its profits and returns by finding the cheapest to produce offshore manufacturing in China and now India.

But do you think there is any chance US firms are going to give up on these rewards of financialisation? Are big firms voluntarily going to cut the cash earmarked for C-Suite bonuses by squandering it on the real economy? Why do you think the tech billionaires, the hedge funders, the money-men and corporate America is so desperate to embrace Trump, Maga and the new Republicans? They are promising to build factories – but let’s see how many actually will, knowing Trump will soon be a yesterday’s issue.

And on that… out of time and back to the day job….

Bill Blain

Author, The Morning Porridge

CEO Windshift Capital

Partner Shard Capital