I’ve built a global screen that highlights small and mid-cap companies globally that are:

• Profitable

• Growing

• Reasonably priced

• Not obviously playing accounting games

The output…

204 stocks that sit in an interesting sweet spot:
too small for big institutions, too boring for social media, and potentially very interesting for a patient individual investor.

Today’s post is about that hunting ground and one stock profile that is a potential buy. The type of setup that can realistically compound at double-digit rates for years.

I’m not giving you a ‘hot tip’ here. I don’t personally believe in investing based on others stock picks as it’s never as simple as it seems. What I am showing you is how I look for asymmetric opportunities, and what will sit behind the deep-dive write-ups I’ll be publishing.

If that sounds useful, feel free to subscribe.

Subscribe now

A universe of choice

I don’t want lottery tickets. I want cash-generating businesses where the numbers already look good before I even open an annual report.

Think about this…

What if you only looked at global companies that are already profitable, growing, and not obviously over-levered but still trade on single-digit / low-double-digit earnings? That’s obviously what we all want right?

Using Koyfin, I built a screen across Canada, Europe GCC and Asia/Pacific, with this base universe:

• Market cap: between $2m and $2bn

• Keeps me in small caps & micro caps while avoiding total illiquid names.

• Volume (notional): at least $100k/day

• Enough liquidity to build a position without moving the price too much.

Then I layered on some quality + value filters:

1. P/E (next 12 months):

   • Between 1x and 12x

   • I’m not paying 25x earnings for a story. If the business is good, I want a margin of safety on entry.

2. Return on Equity (LTM):

   • At least 10%

   • Below that, you’re often just looking at mediocre capital allocation or a cyclical blip.

3. EBIT Margin (LTM):

   • At least 5%

   • I want real pricing power or cost advantage, not paper-thin economics.

4. Revenue growth (3-yr CAGR):

   • At least 5%

   • This isn’t “cigar-butt” investing. I want companies that are still growing the pie, not just cutting costs.

5. Balance sheet cleanliness:

   • Net Debt / EBITDA: between -1x and 2.5x

   • Interest coverage (EBITDA / Interest): at least 3x

   • Enough flexibility to survive shocks, but I’m happy with some sensible leverage.

6. Cash conversion:

   • Cash From Operations (LTM) > 0

   • If earnings are great but cash never shows up, I walk away.

That’s it. No rocket science.

This simple rule-set gave me 204 companies.

When I slice the list by region and sector, it looks roughly like this (numbers rounded):

• GCC and Asia: ~350 names originally, most of the list

• Europe: ~80 names

• Heavy representation in Industrials, Consumer, Energy, Materials, Communication Services

That’s exactly the sort of real-world businesses that never trend on Financial Twitter/X and where mispricings can go unnoticed sit for years.

An example from the screener (small cap Canadian stock. From ~2021-2023)

This is a great example of a business that would put many people to sleep but it returned me 14 times my initial investment.

The company: Data Communications Management (TSX: DCM), a Canadian print and marketing services business.

The numbers from a few years ago looked roughly like this:

• Net income: about C$14m

• Depreciation & amortisation: about C$11m

• Capex: roughly C$1–2m

• Market cap at the time: around C$60–65m

If you stop at headline P/E, you get something like:

P/E ≈ 4–5x (C$60–65m / C$14m)

Which already looks cheap.

DCM was running old print plants that were already built and largely paid for. The accounting was still depreciating those assets every year, but economically, the business wasn’t going to rebuild the same footprint again. In practice, they were squeezing more volume out of existing assets, not constantly replacing them.

If you think in owners’ earnings instead of GAAP earnings, the picture changes:

Owners’ earnings ≈ net income + D&A − maintenance capex ≈ 14 + 11 − ~1 ≈ C$24m

On that basis, the same C$60–65m market cap was trading at roughly:

2.5–2.7x owners’ earnings

Nothing magical happened in the business between 4–5x P/E and ~2.5x owners’ earnings. Same company. Same plants. Same cash flows.

The only thing that changed was how you measure earnings power:

• P/E blindly accepts depreciation as if all those assets will need to be replaced.

• Owners’ earnings recognises that in this case, a big chunk of depreciation was just an accounting exercise of past capex, not a real future cash outlay.

That’s the whole point of using owners’ earnings: you’re trying to answer “how much cash could a rational owner pull out of this business over time?” rather than “what does the accounting standard say this year’s profit is?”.

And that’s why in these situations – heavy depreciation, stable or shrinking asset base, or lots of goodwill amortisation – is where classic P/E and even EV/EBIT systematically understate the true economics of the business.

That’s how you get multi-baggers; earnings growth + a bit of multiple expansion, starting from a low base.

Share

The write up

The screen is just essentially the filter.

For each name that looks like DCM, my next steps are always the same:

1. Understand the business

• What does it actually do?

• Who are the customers?

• Is it selling vitamins or cat food? Nice-to-have or life and death?

Ask yourself: “If this disappeared tomorrow, who would care?”

2. Test the moat and switching costs

Is there a competitive advantage, if so what is it and can it be maintained?

Examples…

• Sticky software / monitoring platforms (think CRM or accounting systems)

• Long-term contracts (multi year binding agreements)

• Products embedded in processes and infrastructure

For each stock on the shortlist, I ask:

• What would it take for a customer to switch to a competitor?

• Are there training, integration, regulatory or reliability costs that make switching a pain?

• Does the business benefit from scale, network effects, or data advantages?

If the answer is “they can easily be swapped out on the next tender”, I’m not interested.

3. Follow the cash and capital allocation

High ROE only matters if:

• It’s sustainable, and

• Management reinvests sensibly.

I’ll dig into:

• Multi-year cash flow statements

• Capex vs depreciation (maintenance vs growth)

• Acquisitions and the returns they’ve generated

• Buybacks and insider ownership

The best situations are where:

• Owner-operators or aligned managers run the show (skin in the game)

• They reinvest in high-return projects first

• And return excess capital via dividends and buybacks.

• Also I look at the ‘little things’ - Ex; during earnings calls is the CEO always present and leading the announcement or have they not even bothered to show up?

4. Price vs quality: what’s already in the multiple?

Finally, I’ll tie everything back to valuation:

• At, say, 9x earnings, what is the market assuming?

• No growth? Permanent cyclicality? Governance risk?

• Do those assumptions match reality, or is there a clear disconnect?

If I can get comfortable that:

1. The business can grow earnings at mid-teens,

2. The balance sheet isn’t a ticking time bomb, and

3. The current multiple doesn’t embed that growth,

…then I’m happy to underwrite a 3–5 year holding period and let compounding do the work.

That’s the point at which it transitions from a watchlist name to a full write-up and, potentially, a position.

What you can expect from my Substack

Over the coming weeks and months, I’ll be providing more ways to find and identify potential multi-baggers. The articles will highlight key pointers and tips on:

• Reading annual reports

• Breaking down unit economics

• Testing the moat and management quality

• Building simple valuation models and scenarios

If that’s the level of depth and transparency you want to see in value investing research, without the institutional jargon, then:

Subscribe now so you don’t miss any future write-ups.

Subscribe now

You’ll see the names, the numbers, and the reasoning and you can decide for yourself whether any of them deserve a place in your portfolio. *Do your own research*

I also write about quantitative trading strategies, AI and other asset classes to invest in.

Thanks for reading.

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

If your entire plan is just to keep a job and hope things work out, you’re basically gambling your whole financial future on something that’s getting shakier by the day.

I’m not saying this as some tech hype person. I’ve spent over 15 years in investment management and consulting, watching how markets and companies actually behave when things change. And here’s what I’ve learned:

As AI becomes more prevalent, your salary might be one of the riskiest bets you’re making.

It’s your investments that will keep you in the game.

Now, when most people think about investing for the AI era, they immediately jump to tech stocks or crypto, hoping to catch the next big thing. Sure, that’s part of it. But it’s not the foundation.

This article is about something way less exciting but way more important: bonds.

To be completely honest with you; Bonds are boring but steady and most likely to be your nest egg if the world falls apart.

For paid subscribers, I go a level deeper, into practical financial valuation modelling, checklists, and real-world case studies drawn from both the markets and the boardroom: how to read a bond like an investor and a consultant, how to evaluate assets, and how to build a portfolio that gives you genuine leverage in the age of AI.

Subscribe now

Your salary isn’t going to cut it

Let’s talk about what AI is actually doing to salaries:

• It’s compressing wages. If AI can handle half your job, companies aren’t going to pay you what they used to.

• It’s polarizing jobs. A small group at the top, people building or running AI-powered operations, do great. Everyone else? Fighting over scraps.

• It’s making everything more uncertain. Whole careers can get reshaped faster than you can adapt.

So here’s the thing: your job isn’t some guaranteed, inflation-beating income stream anymore. Whether you like it or not, you’re exposed to tech disruption.

The question becomes: How do you create income that doesn’t depend on staying employed?

One answer: own stuff that pays you just for holding it.

Stocks and real estate are the usual choices. But there’s another one most people overlook, especially if they’re just starting out: bonds.

What actually is a bond?

Forget the finance jargon. Here’s what a bond really is:

You’re the lender.

You lend money to a government, a company, or some other organization. In return, they promise to:

• Pay you interest (usually at a fixed rate)

• Give your money back at a specific date

While stocks mean you own a piece of a business, bonds mean you’re renting out your money.

Why does this matter when AI is disrupting the world?

Because lending, done carefully, gives you:

• More predictable income

• Less wild ups and downs than stocks

• A way to stabilize your portfolio so you’re not at the mercy of the stock market every single day

Why bonds actually matter right now

Most people think bonds are just for retirees or ultra-conservative investors. That’s not wrong, but it’s partly untrue as well.

In an AI-driven world, bonds do a few really strategic things:

They give you income that isn’t your salary

If your company cuts staff or freezes raises because they find efficiencies with AI, your income drops. Maybe by 100%.

A bond portfolio gives you regular interest payments, a more stable base of cash flow. It’s not about getting rich off bond interest alone. It’s about not being completely dependent on one source of income - a salary.

They keep you stable when everything else falls apart

AI is going to create massive winners and losers in stocks and crypto. That means wild swings, narratives that flip overnight, and brutal losses if you’re betting on the wrong story.

Quality bonds, especially government and solid corporate ones, tend to move differently than stocks. When stocks are tanking, bonds often hold steady or even go up. That’s diversification in action.

It means you’re less likely to panic and sell everything at the worst possible moment.

Share

They give you dry powder for opportunities

Here’s a key benefit: bonds act as ammunition.

If you’ve got a chunk of your portfolio in stable bonds and some great AI stock crashes 40% in a panic, you have:

• Something stable to sell

• The ability to buy into the chaos

• The mental space to be greedy when everyone else is fearful

Without something stable, you’re just another person watching their portfolio bleed and hoping it comes back someday.

The main types

There are tons of bond types, but you don’t need a finance degree like me. Here’s what matters:

Government bonds

You’re lending to a government. Generally lower risk, especially in stable countries. Different lengths available; short-term, medium, long-term. This is usually where beginners start.

Investment-Grade corporate bonds

You’re lending to big, relatively stable companies. They are rated riskier than government bonds, so they pay more interest. Still considered safer in the corporate world. Often issued by companies you’ve actually heard of.

High-Yield (junk) bonds

You’re lending to financially weaker companies. They pay higher interest because they’re riskier. They can do great in good times and blow up in bad times.

For beginners, these aren’t the foundation. Maybe a small piece later if you know what you’re doing.

What you need to understand (the important stuff)

Know these core concepts:

Maturity

When the bond pays you back your original money. Shorter maturity = less uncertainty, usually lower returns. Longer maturity = more uncertainty, usually higher returns.

In this AI-disrupted world, shorter or intermediate maturities often make more senseless exposure to long-term unknowns.

Coupon and Yield

The coupon is the fixed interest rate. The yield is what you actually earn based on what you paid for the bond. When bond prices fall, yields rise (and vice versa). You don’t need to memorize formulas, just get that price and yield move in opposite directions.

Credit risk

The risk they won’t pay you back. Higher credit risk = higher interest, but also higher chance of pain. In a chaotic AI economy, stick with stronger credit rated bonds for your core holdings rather than chasing every extra bit of yield.

Interest rate risk

When interest rates rise, existing bonds become less attractive (new ones pay more), so their prices usually fall. This hits longer bonds harder. Just remember: longer bonds = more sensitive to rate changes.

How to actually buy bonds

You don’t need to call some bond trader at a bank. Regular investors have real options now:

Bond ETFs and mutual funds

For most beginners, this is the way. You buy a fund that holds a basket of bonds. Instant diversification. Trades like a stock on your brokerage platform.

You can find funds for government bonds, corporate bonds, different maturity ranges; whatever fits your goals.

Individual bonds

You buy specific bonds directly. You know exactly what you own and when it matures. You can build a “ladder” with bonds maturing in different years.

Downsides: often higher minimums, more research needed, less diversification unless you have serious money. This makes sense later, but not essential starting out.

Bond ladders

A ladder is just a series of bonds with staggered maturity dates (like 1, 2, 3, 4, 5 years). Benefits: you always have something close to maturing, it reduces interest rate risk, and you get regular opportunities to reinvest or grab new opportunities.

A framework that actually makes sense

This isn’t financial advice, but here’s a way to think about it:

Figure out why you want bonds. Stability? Income? Dry powder for future moves?

Start with quality. Government and investment-grade corporate bond funds as your foundation. Keep it simple before reaching for higher yields.

Mix your maturities. Some shorter-term stuff for flexibility. Some intermediate-term for better yields without going crazy long.

Size it based on your other risks. Taking big swings elsewhere (tech stocks, crypto)? A meaningful bond allocation makes sense to balance things. More conservative overall? Bonds might be a bigger chunk.

Automate if you can. Small, regular contributions beat waiting for “the perfect moment” that never comes.

Don’t make these mistakes

I’ve seen these over and over:

Chasing yield blindly. Buying whatever pays the highest without understanding why. High yields exist for a reason, usually risk.

Ignoring interest rate risk. Loading up on long-duration bonds without realising how vulnerable they are when rates move.

Treating bonds as dead money. Dismissing them because they’re boring. Your goal isn’t excitement, it’s resilience.

No clear purpose. Owning bonds just because someone said you should, without knowing if they’re for income, stability, or optionality.

Bonds aren’t the core of your portfolio. They’re the reduced risk allocation that keep everything functional.

The bigger picture

Look, bonds alone won’t make you financially free.

But in the age of AI, your job is riskier, your income path is more fragile, and if you’re only chasing high-growth assets, you might be exposed to some wild swings.

Bonds give you:

• Income that isn’t your salary

• Stability when everything else is going haywire

• Ammunition to grab opportunities when AI panic sends good assets crashing

While everyone’s obsessing over the next AI stock or crypto token, building a solid fixed-income foundation in the background might be one of the smartest, most underrated moves you can make.

One last thing

I spend my time thinking about the intersection of AI, the future of work, financial markets, and how actual businesses make money.

My goal is simple: help you shift from being exposed to AI as an employee to being positioned for AI as an asset owner.

If this resonated with you and you want to go deeper; specific bond strategies, how to build an AI-resilient portfolio, real company case studies through financial modelling, then consider subscribing. And if you’re serious about this stuff, the paid tier is where I go into detailed frameworks and examples.

Your job might not survive the AI age. But your assets? If you choose them well, they give you options either way.

Thanks for reading.

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

Do you know what protects Bitcoin? Mathematics.

From what I have researched the future of Bitcoin will be compromised. What I’ve found is both elegant and terrifying.

The incredible simplicity of Bitcoin’s armour

Bitcoin’s security doesn’t actually rest on SHA-256, the hashing function most people associate with cryptography (crypto). That’s not where the real protection lies. The real armour or ‘Achilles heel’, depending on your perspective is an elliptic curve with a deceptively simple equation:

y² = x³ + 7

That’s it. High school algebra. Yet this mathematical object, specifically the curve called SECP256K1, protects every bitcoin in existence.

SECP256K1 breaks down like this: SEC stands for Standards for Efficient Cryptography, P means prime field, 256 refers to the 256-bit key size, K indicates it’s a Koblitz curve (named after one of the mathematicians who pioneered elliptic curve cryptography), and the “1” just distinguishes it from other curves in the family.

What makes elliptic curves effective is their one-way function property. You can easily compute an output from an input, but reversing the process, i.e. going from output back to input, is computationally infeasible. It’s like mixing paint: combine blue and yellow, you get green. But green paint cannot extract the original blue and yellow.

This is why nobody can steal Bitcoin even with 10 million NVIDIA GPUs at their disposal.

At least, not yet.

Mix AI with math problems

By treating this research piece as a problem statement I thought about the security itself as a potential ticking time bomb then something interesting happened. AI started getting better at mathematics.

Most people don’t realise how hard math actually is. I mean math is really hard. Some problems have taken 400 years to solve. The Riemann Hypothesis has been unsolved since 1859, over 160 years, and might never be cracked. Math puzzles can persist for centuries, sometimes millennia.

But AI is getting smarter and faster. Companies like OpenAI are working on mathematical reasoning. Vlad Tenev has a math AI startup. Google DeepMind is tackling the Clay Millennium Problems; seven famous unsolved problems with a million-dollar prize on each.

Elliptic curves aren’t part of the Clay prizes. They’re the trillion-dollar prize. Because everyones Bitcoin is guarded by them.

Most mathematicians believe there’s no analytical solution to the elliptic curve discrete logarithm problem. I actually think they’re wrong. I believe an elegant, easy solution exists, we just don’t know it yet. It might take hundreds of years for humans to discover it. Or it might take six months of AI working through the math to prove it.

The classical attack that exists today is called Pollard’s rho algorithm. It guarantees finding the answer in roughly √n operations, where n is the size of the key space. For Bitcoin, n = 2^256, so √n = 2^128 operations.

There aren’t enough computers on Earth to do that. But my research suggests it’s not as far off as most people think. With careful math, some incremental improvements (not breakthroughs, just steady progress), and Moore’s Law running for another decade, we might see a viable classical attack combining analytical insights with massive computational power.

A quantum of time

Before that happens, though, most experts expect quantum computing to break Bitcoin first.

Quantum computers can run Shor’s algorithm, which efficiently factors large numbers and solves the discrete logarithm problem; exactly what protects Bitcoin. Running Shor’s algorithm on SECP256K1 requires somewhere around 2 million qubits, possibly fewer with optimization. We don’t even have one stable logical qubit yet, but progress is accelerating.

Some companies claim we’re five to ten years away from quantum computers that can break Bitcoin. Others say thirty to fifty years. IonQ has suggested it could happen in a year, though nobody credible believes that timeline.

What you can’t argue, and I’ve never met anyone credible who does, is that it will never happen. Even the most conservative quantum skeptics admit that eventually, someone will build a quantum computer capable of running Shor’s algorithm.

And here’s where Bitcoin’s security promise as ‘digital gold’ becomes problematic. If Bitcoin is supposed to be hard money that lasts forever, fifty years from now matters. Thirty years from now definitely matters. Ten years from now is a crisis.

Yet the Bitcoin price hasn’t dropped. The market is telling us it’s not worried. Why is that?

Share

The fork problem

Technically, Bitcoin can be saved. You can fork the code and upgrade to quantum-resistant cryptography. The technology exists; post-quantum cryptography (PQC) algorithms are already developed, freely available, and relatively easy to implement.

But there are two massive problems.

1: The Bitcoin community hates change. The resistance to upgrading Bitcoin’s core protocol is analogous to religious resistance to change. Despite this, Bitcoin has undergone six major upgrades over its history, so it’s not impossible. If forced by an existential threat, the community would likely upgrade.

2: What do you do with Satoshi’s coins?

Let’s say Bitcoin forks to a quantum-safe protocol. Holder are told to move their tokens to the new protocol by a given date, and they’ll be protected. If everyone moves their tokens save a few how are no longer with us..

E.g. What about Satoshi’s coins? Most people believe Satoshi is dead. Those coins aren’t moving as are anyone else’s that isn’t alive or able to move their tokens.

If you leave old addresses active, the first person with a working quantum computer steals Satoshi’s (and other dead folks that are still holding BTC) stash. It’s roughly a million Bitcoin for Satoshi plus whatever the number is for other people no longer with us. These tokens can, potentially, be moved to a quantum-safe wallet. Once quantum computers exist, everyone’s private keys become public. The system is worthless.

So you have to force everyone off the old protocol. Set a drop-dead date: after this date, old addresses can’t move coins. However this won’t solve the problem…

Every Bitcoin that hasn’t been moved by the deadline e.g. Satoshi’s BTC, lost wallets, forgotten keys effectively gets destroyed. Billions of dollars gone. Some countries handled currency transitions this way when adopting the Euro. Germany said you had until 2002 to exchange Deutsche Marks, after which they became worthless.

This solution works, but it fundamentally undermines Bitcoin’s value proposition. It’s no longer immutable hard money. It’s money that needs periodic emergency upgrades with arbitrary deadlines, each one potentially destroying wealth.

Even gold had this problem. Americans couldn’t legally own gold during the Bretton Woods era. But at least gold doesn’t have a cryptographic expiration date due to its physical nature.

The BTQ debate

BTQ, a company that’s supposed to solve these problems and is currently valued at roughly a billion dollars.

Let me walk through what they’ve built with $41 million in total funding and approximately $2.5 million in annual R&D spend.

BTQ claims to offer quantum-secure solutions across several areas: stable coin networks, defense applications, and quantum computing hardware. Their pitch deck says they’re the “first company to achieve quantum advantage” and can “outperform deca billion dollar competitors with minimal funding.”

These are interesting claims for a company with effectively zero revenue.

Their quantum hardware slide misspells “neutral atom” as “neutural atom.” Their chip design uses 65-nanometer TSMC nodes; technology from roughly 2005, ancient by semiconductor standards. They claim to have “produced tens of millions of chips deployed within LG,” but there’s no corresponding revenue on the books.

Their main business proposition seems to be helping companies transition to post-quantum cryptography. The problem? is that PQC algorithms are freely available and designed to be easy to implement. They’re open source. Thousands of companies have already deployed them. Why would anyone pay BTQ for this?

The stable coin angle is equally perplexing. Everyone wants to build a stable coin. Tether and Circle are valuable because they keep the yield from deposits, you hold the dollar equivalent, they earn interest on billions. But launching a new stable coin requires getting people to actually use it. BTQ’s stable coin won’t run on Ethereum or Solana or any established network. It’s its own thing, which means building an entirely new ecosystem from scratch.

Compare BTQ’s billion-dollar valuation to actual companies at the same level: Docker, Quora, Patreon. These companies have real revenue, established products, millions of users. BTQ has PowerPoint slides and aspirations.

I’ve done startups. I get it. Early stage companies are mostly business plans. But I don’t ask for a billion-dollar valuation until I have $100-200 million in revenue. BTQ is asking for unicorn pricing at the pre-seed stage.

What this all means

The real story here isn’t BTQ’s questionable valuation. It’s the gap between Bitcoin’s vulnerability and market awareness.

Whether through quantum computing or classical improvements, Bitcoin faces a crypto crisis within decades. The market shows zero concern. Bitcoin’s price reflects no discount for this existential risk.

Maybe the market is right. Maybe quantum computing is further away than even conservative estimates suggest. Maybe some brilliant solution emerges that lets Bitcoin upgrade seamlessly without destroying dormant coins.

Or maybe the market is dramatically mispricing the risk because the math is too hard, the timeline too uncertain, and the problem too abstract.

After spending a few years studying elliptic curves, working through the mathematics, understanding the attack vectors. I gained serious respect for the difficulty of these problems. I also came away convinced that Bitcoin’s crypto foundation is more fragile than most people realize.

Satoshi made a choice in 2008: 256-bit security instead of 512-bit. It seemed reasonable then. Moore’s Law wasn’t as obvious. Quantum computing was theoretical. AI capabilities were primitive.

But 256-bit security in 2025, let alone 2035 or 2045, looks increasingly inadequate. The math hasn’t changed however the threat landscape has.

Bitcoin is supposed to be digital gold, money that lasts forever. But forever is a long time. And the crypto walls protecting Bitcoin’s trillion-dollar market cap are flimsier than most people think.

The question isn’t whether Bitcoin can be broken. The question is when and whether anyone will notice in time to do something about it.

Thanks for reading.

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

That might sound dramatic. But it’s not clickbait.

Businesses are relentless in terms of cost savings. If you follow how markets react to new information and new technology, it’s just where the logic leads.

I’ll walk you through it.

From efficient markets to efficient AI

Eugene F. Fama defined a market to be “informationally efficient” if prices at each moment incorporate all available information about future values.

To put it simply:

• Markets process new information fast

• Prices adjust

• And over time, it becomes very hard for any one investor to consistently “beat” the market

• On average 8/10 companies are fairly valued on this basis.

Now take that same logic and point it at labour markets instead of financial markets.

What happens when companies can see, very clearly and very quickly, what AI can do, what it costs, and how it performs versus humans?

Fair value pricing in financial markets hinges on a ruthless equation: lower costs equal higher earnings. For companies, this means AI-powered automation isn’t just an option; it’s a competitive imperative.

Every role that can be replaced by machines represents a permanent cost reduction, higher margins, and increased shareholder returns. The incentives couldn’t be clearer, and they all point in the same direction: away from human labor.

That’s where something I’ve been thinking about comes in: The Efficient AI Hypothesis (EAIH)

In plain English:

As AI gets better, employment markets become ruthlessly efficient at allocating tasks to the cheapest, most scalable form of intelligence. Over time, that’s not human beings. It’s machines.

Here’s the framework.

The efficient AI hypothesis (EAIH): What I mean by it

Efficient AI hypothesis (EAIH):
As AI and automation spread, labour markets become increasingly efficient at matching tasks to the cheapest and most scalable form of intelligence. Over time, a growing share of economic tasks are completed by AI, and the structural demand for human labour falls.

Three forces drive this.

1. Information

Companies will know, almost in real time:

• What AI systems can do

• How reliably they do it

• And how much it costs

Your job description stops being “Marketing Manager” or “Analyst” and turns into a list of tasks:

• Write X

• Analyse Y

• Prepare Z

Each task gets a simple question attached to it: human or machine?

2. Friction

Hiring a human is expensive friction:

• Recruitment

• Interviews

• Contracts

• Onboarding

• HR, compliance, management

Turning on an AI agent?

• Log in

• Add credit card

• Paste API key

That’s it.

The friction advantage is completely on the side of the machine.

3. Price

Once AI can do a task at nearly zero marginal cost, that becomes the benchmark.

If you’re a human doing the same task, you have to justify why you’re worth that premium. If you can’t, the machine wins.

Under EMH, alpha (excess returns) gets competed away.
Under EAIH, employment arbitrage (being paid more than the machine alternative) gets competed away.

Same logic. Different battlefield.

Weak, semi-strong, and strong EAIH

Shamefully borrowing from EMH, you can think of EAIH in stages.

Weak EAIH – We’re already living this

AI discreetly automates low-skill, repetitive, process-driven work:

• Data entry

• Basic customer support

• Simple copywriting and content

• Routine back-office operations

This isn’t the future. This is now.

Semi-strong EAIH – This is where it gets concerning

AI moves up the value chain and starts eating into “respectable”, credentialed work:

• Junior analysts and associates

• Consultants building slides and models

• Developers writing code

• Paralegals and legal researchers

• Marketing, PPC, SEO, campaign optimisation

The important distinction: AI doesn’t instantly delete all these jobs.

Instead, it compresses the pyramid.

You still need seniors. You still need decision-makers. But:

• One senior with the right AI tools

• Can do what previously needed a small team of juniors

So the org chart doesn’t evaporate. It just shrinks from the bottom up.

Strong EAIH – The endgame

This is where AI systems and autonomous agents run most of the economic machine:

• AI runs simulations, pricing, and optimisation

• AI manages supply chains and logistics with minimal human oversight

• AI helps design, test, and iterate new products at speed

Companies start to look like:

• Intellectual property

• Data

• Distribution

• A very small group of humans on top

Humans don’t vanish—but the humans-to-output ratio collapses.

How it actually unfolds (it’s not a Sci-Fi movie)

EAIH doesn’t show up with a robot dragging your chair out of the office.

It’s much more subtle than that.

Subscribe now

1. AI eats tasks before it eats jobs

No CEO wakes up and says, “Let’s fire 40% of the company today.”

They start with what can be automated.

• Break roles into tasks

• Highlight anything repeatable, rules-based, or information-heavy

• Deploy AI to handle 20%, then 40%, then 60% of those tasks

The result:

• One person now does the work of two or three

• Hiring slows down

• Teams look “leaner” and “more productive”

The changes are felt in the job market.

2. Internal productivity = External job destruction

Inside the company:

• Costs per unit of output fall

• Margins improve, especially for early adopters

Outside the company:

• Competitors have to copy or watch their earnings get squeezed

• Entire industries recalibrate what “normal” headcount looks like

At the sector level:

• Fewer people are needed to produce the same output

• New roles don’t grow fast enough to absorb displaced workers

This is how you get that creeping feeling of ‘it’s getting harder to get in and harder to move up.’

3. AI as borderless digital labour

AI doesn’t care about:

• Visas

• Time zones

• Sick days

• Office politics

• Annual reviews and pay rises

It just runs.

For a business, the decision becomes painfully simple:

Hire a human for $50k–$150k a year
or
Pay a fraction of that for AI tools and compute

Once those AI-driven workflows are stable and trusted, the pressure to reduce headcount is inevitable.

The great trap: “Don’t worry, AI will create new jobs”

This is the line you hear everywhere:

“Yes, some jobs will disappear, but new jobs will be created…. prompt engineers, AI trainers, implementation consultants…”

There is some truth there.

But it completely misses the dynamic I’m talking about with EAIH.

We are seeing a boom in what I’d call transient AI support roles:

• Prompt engineers

• AI product trainers

• Data labellers and annotators

• “Human in the loop” QA

• AI workflow designers

• Implementation / integration consultants

These are real roles. They pay real money. They’re amazing short-term opportunity sets if you get in early.

But look at the pattern.

Phase 1 – Early

• Everything is messy and ad-hoc

• Companies need humans to experiment, stitch things together, and put basic guardrails in place

• Pay looks great because very few people can do it properly

Phase 2 – Standardisation

• Best practices emerge

• Playbooks, templates, no-code tools, off-the-shelf solutions appear

• Prompt engineering stops being a mystic art and becomes a feature baked into products

• AI systems start tuning themselves and optimising their own inputs and outputs

Phase 3 – Automation of the Support Layer

• The human ‘glue’ that sits between business and AI becomes the next target for automation

• Prompting, evaluation, data labelling, and QA increasingly become AI-to-AI workflows

• The arbitrage, being paid a premium to sit on top of AI, gets competed away

So yes, these jobs are real. But they are bridges, not destinations.

Under EAIH, if your role exists primarily to help AI function better, it will eventually be in AI’s sights as well.

Businesses will harbour zero loyalty - it’s in their track record

When manufacturing was the main engine of growth, rich countries like the US and those in Europe enjoyed rising GDP per capita and the comfort of ‘home-made’ goods and services. The catch was simple: labour was expensive.

Then China, India and a wave of emerging markets came along with cheaper, scalable human capital across the board; manufacturing, IT, back-office, customer support. Offshoring went from ‘cost-saving experiment’ to do-or-die strategy for corporates and even smaller firms that wanted to stay competitive and protect their margins.

Did that create employment gaps back home? Yes, especially in specific sectors and regions. The academic evidence is nuanced at the macro level, but it’s very clear at the local level: moving production and tasks abroad has meant job losses, higher unemployment and wage pressure for workers in exposed industries. Studies1 on offshoring and trade show that relocating production can trigger job destruction and periods of unemployment for affected workers, and the ‘China shock’ literature links rising imports from China to large drops in manufacturing employment and weaker labour market outcomes in the regions that competed directly with those imports.

AI is simply the next chapter of the same story.

Instead of offshoring tasks to cheaper humans, companies will offload tasks to AI systems that are plugged into their entire digital infrastructure; CRMs, ERPs, data warehouses, support tools, everything. The output is cheap, fast and infinitely scalable. And this won’t just be an enterprise story. Small and medium-sized businesses, which live or die by cost discipline, will be highly motivated to automate aggressively just to survive, let alone grow.

The pattern is the same: when there’s a cheaper, more efficient way to get the work done, businesses take it. Not because they’re evil. Because that’s their track record.

Why UBI probably won’t arrive in time

I broadly agree with the likes of Frank Giustra on this:

• AI will concentrate wealth and productivity in fewer hands

• Disruption won’t stop at low-skill jobs; it will move deep into white-collar territory

• ‘Just re-skill’ massively underestimates the speed and scale of what’s coming

When you look at it realistically, some form of Universal Basic Income (UBI) or income floor starts to look less like a utopian dream and more like a social safety valve.

Without something like that, you’re staring at:

• Rising social unrest and political volatility

• Populist policies that might actively damage innovation and capital

• Collapsing consumer demand as too many people lose stable income

Here’s the issue: policy is slow. Technology is not.

Even if UBI or UBI-like programmes eventually arrive, they will almost certainly lag the disruption.

You cannot build your personal plan on the assumption that government will save you on time.

So I’d suggest operating as if:

• EAIH is roughly right

• And you are responsible for your own safety net.

Your salary is a single, concentrated bet

Think about your job the way a portfolio manager would:

Most people have:

• One main employer (or one small business)

• One primary skillset

• In one country, under one regulatory regime

That is the equivalent of putting almost your entire net worth into one stock and holding it forever while the underlying business is being disrupted.

Under EAIH, your job income is exposed to:

• Technology risk – your tasks are being benchmarked against AI constantly

• Management risk – your leadership can decide, at any point, to automate, outsource, or restructure

• Macro risk – recessions and downturns accelerate automation because cost savings suddenly become urgent

Upside of your salary?

• A pay rise once a year if you’re lucky.

Downside?

• Lose that one income stream, and everything gets very real, very fast.

That’s why I keep coming back to this:

In an AI-driven world, your human capital, your ability to sell your time, is one of the riskiest assets you own.

So who actually wins?

If EAIH is the direction of travel, who comes out ahead?

Not the average worker whose job is getting hollowed out.

The winners are:

• Businesses that use AI aggressively and intelligently

• Owners of those businesses; shareholders, equity holders, partners

• Owners of real assets and infrastructure that those AI-powered companies depend on

As AI improves:

• Margins expand (at least for a while)

• Productivity per employee explodes

• You need fewer people to generate bigger profits

And here’s the key piece that ties back to EMH:

Markets are not stupid.

Any ‘free’ profitability created by AI gets baked into the share price fairly quickly:

• The more obvious and scalable the AI advantage, the faster it gets priced in

• Early, informed capital can benefit

• Late, uninformed capital gets average (or worse) returns

You end up with a double squeeze:

• Labour side: employment arbitrage gets competed away as AI takes more tasks

• Capital side: profitability arbitrage gets competed away as markets price in AI gains

This doesn’t mean there’s no point investing. It means:

• You can’t rely on labour income alone

• You can’t just buy “AI stocks” blindly and hope for the best

• You need a framework to separate hype from real cash flows

The only real hedge: Stop being just a worker

The uncomfortable but honest takeaway:

You need to move from being only a worker to being a worker and an owner.

That doesn’t mean quitting your job tomorrow. It means gradually building exposure to:

• Equities – ownership in businesses that actually benefit from AI (not just mention it on earnings calls)

• Bonds / fixed income – cash flows and some stability

• Real estate/REITs – when it makes sense, as long-term, income-producing assets

• Commodities and real assets – as hedges against currency and contagion risk

• Selective asymmetric bets (including parts of crypto) – only if you understand the risks and position size properly

This is not about day trading, leverage, or chasing whatever is trending on X this week.

It’s about building an AI-resilient balance sheet over the next 5, 10, 20 years:

• So if your job disappears or your industry gets re-priced, you’re not starting from zero

• So you share in the upside of the very systems that are making labour markets more efficient and more brutal.

What I’m doing on this Substack (and why it exists)

The reason I’m writing about this is simple:

The usual advice; '“just learn to code”, “just adapt”, “just re-skill”, is nowhere near enough for what’s coming.

Here’s what I’m trying to do with this Substack:

• Put structure around ideas like EAIH
  So you’ve got a clear mental model, not just headlines.

• Break down how AI is actually changing industries
  Where it boosts margins, where it destroys pricing power, and who actually gets paid.

• Teach an investor’s way of thinking about:

  • Income vs assets

  • Risk vs resilience

  • Valuations, cashflows, and management quality in an AI world

If you’re a free subscriber

You’ll get:

• Big-picture posts

If you join as a paid subscriber

That’s where I go deeper and more practical:

• Industry and company breakdowns through the lens of EAIH

• How to think about stocks, bonds, and other assets in this new environment

• Live and pre-recorded company valuation modelling (this is how investment management professionals make investing decisions)

• Portfolio construction ideas

• Ongoing updates as AI, markets and policy shift, because this is not a static situation

I’m not promising certainty. No one has that.

What I’m trying to give you is clarity and a structured way to secure your future.

You’re not going to beat the machines at being machines

EMH made it hard for stock pickers to beat the market.
EAIH will make it hard for most people to “beat” the machines in the labour market.

You don’t need to:

• Predict the exact year AGI arrives

• Pick the winning AI model

• Or have hot takes on every new model announcement

But you do need to stop assuming that:

• One job is a safe, permanent asset

• AI is just another tech buzzword you can ignore until retirement

If you want to stay ahead of the Efficient AI Hypothesis instead of becoming another datapoint in it, you need to start making strides:

• From only human capital → to human capital + financial capital

• From being replaceable labour → to owning pieces of the systems doing the replacing

Ready to start building an AI-resilient future?

If this resonates, and you want to turn these ideas into an actual plan rather than just a worrying read, here’s what I’d suggest:

• Subscribe to this Substack (@usmanconsulting) if you haven’t already

• Stay for the free posts if that’s where you’re at

• Upgrade to paid if you want to go deeper into:

  • Specific companies and sectors

  • Financial modelling (the investment management way)

  • Portfolio thinking in an AI world

  • Practical ways to position yourself on the winning side of EAIH

We’re not going to stop AI from making markets more efficient.

The choice we do have is which side of that efficiency we sit on.

Wishing you all the best,

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

Subscribe now

If you zoom out on today’s AI market, it doesn’t look like a normal industry.

We have however seen similarities before:

Japanese keiretsu

• Cross-shareholdings between banks and industrial firms (e.g. Mitsubishi, Mitsui and Sumitomo).

• Profits and capital recycled inside the group; outsiders struggled to compete or influence governance.

Korean chaebol

• Complex ownership webs (e.g. Samsung, Hyundai, LG).

• Internal trading and shared control, later targeted by regulators and governance reforms.

Dot-com telecom boom & vendor financing

• Equipment vendors financing their own customers to buy more kit.

• Looked great as long as demand rose; when it stalled, defaults and write-downs blew the loop apart.

The ‘AI Money Machine’ has similiar vibes with all of the above; a tight network of companies supplying each other, funding each other, and reinforcing the cycle.

At the centre sits Nvidia, selling GPUs to everyone from Microsoft and OpenAI to obscure AI cloud startups you’ve never heard of. Those same companies sign multibillion-dollar cloud and chip deals, build models that need even more GPUs, and then invest back into the ecosystem that’s feeding them. Capital goes in one side, chips and models come out the other, and for now everyone looks richer.

Now we’ve added a new twist: those GPUs are no longer just hardware. They’re collateral.

Startups are borrowing billions against racks of Nvidia H100s and Blackwells, packaging chips and their future rental income into structured credit deals and ABS structures that look suspiciously like something we’ve seen before in other cycles. On the way up, that leverage amplifies the boom. On the way down, it’s often what turns a story from cyclical slowdown into violent unwind.

This post isn’t about predicting the exact top in Nvidia or calling the end of AI. It’s about something more basic: recognising the structure you’re looking at, understanding how it has behaved in past cycles, and thinking about what it means for you as an investor.

GPUs as collateral. When hardware becomes a credit instrument

For most investors, a GPU is still just a very expensive graphics card.

For lenders, it’s becoming a financial asset they can underwrite, margin and securitise.

Here’s what’s happening:

• AI cloud startups need vast amounts of compute to compete. Buying thousands of H100 or Blackwell GPUs up front is brutally capital-intensive.

• Equity alone won’t get them there, and they don’t want to dilute at today’s valuations if they can avoid it.

• This brings in private credit funds, banks and alternative asset managers. They offer multi-billion-dollar facilities secured by the GPUs themselves and the cash flows from renting them out.

• In some cases, those loans are wrapped in SPVs and turned into asset-backed securities or chip-based ABS.

Economically, this does two things.

First, it pulls demand forward. Instead of buying GPUs only as fast as their cash flows allow, these companies can lever up and buy far more capacity today on the assumption that future AI workloads will fill the racks. That’s fantastic for Nvidia’s order book and pricing power in the short term.

Second, it reframes the risk. The bet is no longer only ‘will AI be big?’ but also:

1. Will these specific GPUs hold their value long enough to pay the debt down?

2. Will utilisation and pricing stay strong enough to service double-digit borrowing costs?

3. What happens if a new architecture, a more efficient chip, or a pause in AI spending makes this generation of hardware less economic to run?

As long as everything is going up and to the right, nobody worries. Collateral looks great, secondary prices are strong, and the deals amortise nicely. The structure feels less risky.

Share

The stress scenario is where it gets interesting.

Imagine a world where AI spend doesn’t collapse, but simply slows down:

• Hyperscalers (AWS, GCP and Azure) decide to sweat existing capacity a bit longer.

• Enterprises drag their feet on rolling out AI applications beyond pilots.

• A more efficient generation of chips appears faster than expected.

Suddenly, utilisation on some of these leased GPU clusters drops, pricing softens, and cash-flow projections start to miss. At the same time, the resale value of the older GPUs gets marked down as the market anticipates the next generation.

For a highly levered AI cloud, this is where the credit documents start to matter. Lower cash flows and lower appraised collateral values push loan-to-value ratios higher. Covenants get tight. Lenders become nervous. Some facilities get amended and extended. Others don’t.

If a few players fail, lenders can and will exercise their security interests, seize the GPUs, and try to sell them. A handful of distressed auctions in an illiquid secondary market can move prices quickly. Lower prices feed back into the models for everyone else’s collateral. Lending committees notice. New GPU-backed lending slows or gets repriced higher. The easy money that was once financing Nvidia’s growth engine starts to evaporate.

At that point, the ‘AI money machine’ looks less like an elegant web of strategic partnerships and more like what it has always been: a leveraged capital structure tied to a single critical component: GPUs.

So what’s next?

AI is real. Large models, agentic systems and AI-native applications are not going away. Nvidia has earned its position at the centre of the stack through genuine engineering excellence and execution.

But markets don’t just price on sentiment. They price on value.

Historically, circular webs of companies that finance, supply and invest in each other can look bulletproof on the way up: Japanese keiretsu, Korean chaebol, telecom vendor financing, even the early ecosystem plays of the last tech cycle. They all worked beautifully until a familiar combination arrived:

• a slowdown in end-demand,

• too much leverage sitting on the same assets, and

• outside capital deciding it no longer wanted to subsidise the loop.

The collateralisation of GPUs is the leverage layer most people never see. It doesn’t mean the whole AI complex is doomed; it does mean that when the cycle turns, the move from soft landing to hard reset can happen faster than most equity investors expect.

The markets tend to stair step higher and then take the elevator down.

If you’ve read this far, you’re already ahead of the crowd that only sees the quarterly earnings headlines and not the credit plumbing underneath.

In the paid edition of this newsletter, I’m going to go deeper into the practical side of this:

• how to read financial reports and analyse by building models to spot where risk and potential return is really sitting in AI & other companies,

• the checklist I use to separate real compounding businesses from good stories on fragile balance sheets, and

• how I think about position sizing and scenario analysis around investments like Nvidia, the broader AI infrastructure theme and other sectors.

If you want to build your own framework for navigating the market rather then join the paid tier and follow along.

Wishing you all the best,

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

Subscribe now

In contrast, the net worth of some of the wealthiest individuals on the planet can be correlated to sizeable equity holdings that have increased over time from investments they have made or the businesses they own (e.g. Warren Buffett / Elon Musk respectively).

If AI is going to make certain companies more efficient, more profitable, and more dominant, then the real winners will be the people who own those businesses, not just those who work for them. This article, and the wider purpose of this Substack, are about helping you shift from being a worker to becoming an informed owner of companies: someone who understands how to build wealth through owning stocks in a deliberate and intelligent way.

I come at this from a unique angle. I’ve spent 15+ years in finance and investment management, analysing companies through their financial statements, valuations, cash flows, and capital structures. Now I run a management consulting business, working directly with leadership teams on pricing, operations, strategy, and all the messy human issues that never show up cleanly in an Excel model. In other words, I see both sides: how markets price a company, and how that company actually creates or destroys value in the real world.

That blend is what I’m bringing to you here. In the free posts, you’ll get clear explanations of how to think about investing in an AI-driven world: why owning stocks matters, how different markets and asset classes behave, and how to avoid the most common mistakes.

For paid subscribers, I go a level deeper, into practical financial valuation modelling, checklists, and real-world case studies drawn from both the markets and the boardroom: how to read a business like an investor and a consultant, how to evaluate management teams, and how to build a portfolio that gives you genuine leverage in the age of AI.

Subscribe now

If you’re serious about not becoming financially obsolete as technology advances, and you want guidance from someone who’s sat on both sides of the table, then this publication is for you. Let’s start by unpacking why owning stocks is no longer optional if you want to truly get ahead in the age of AI.

Below I provide a breakdown showing how to approach investing in the stock markets.

Build a checklist

The best way to do this is by drawing up a checklist that is unique to your investment criteria and risk tolerance. E.g. you only invest in companies that you can understand well, think Coca-Cola or McDonalds. Avoid investing in companies that you do not understand, e.g. precious metals miners or biotech. Next, you define the range in which you want a company to be in. E.g. P/E < 10, market capitalisation of $300m to $1b etc. Paid subscribers have access my main checklist.

Subscribe now

Which markets to invest in

Whilst most of the world invests in the US markets I would encourage you to look at emerging markets as well. The reasons for this are that most of the US market is ‘fairly valued’ (making any perceived alpha much harder to find) and increasingly driven by passive flows into index funds and ETFs rather than true price discovery.

In contrast, many emerging markets are still under-owned, under-researched, and often mispriced. You tend to see greater dispersion between good and bad businesses, which is exactly where thoughtful investors can find opportunity. Yes, the headlines will talk about higher political risk, weaker currencies, or less transparency and those risks are real. But they also mean that when you do the work to understand a country, its institutions, and its leading companies, you’re not competing with a million Robinhood traders and half of Wall Street for the same idea.

That doesn’t mean ignoring the US altogether. It means building a global portfolio where the US provides stability and depth, while select emerging markets (think GCC, India, parts of LatAm, etc.) provide growth, mispricing, and the potential to generate genuine alpha over the long term. If you’d like full access to the stocks I research and buy across US and emerging markets, please check out my paid subscription.

Subscribe now

Know your circle of competence

Investing in the financial markets is, at its core, more about knowing yourself than anything else. If you’re not yet confident in your ability to invest, the solution isn’t to avoid markets altogether, it’s to learn, start small, and deliberately build your skillset over time. One of the most common reasons people lose money is simple: they don’t truly understand what they’re buying. They follow tips, chase headlines, copy what others are doing, and then panic when prices move against them, locking in losses that were avoidable.

The key to avoiding this is understanding your circle of competence. This is the set of industries, business models, and situations you genuinely understand not just at a surface level, but well enough to explain to someone else in plain language. Inside that circle, your judgment tends to be clearer. Outside it, your error rate goes up dramatically.

Your circle of competence might start very small: maybe you understand consumer apps, real estate in your city, or banks in your home country. That’s fine. The goal isn’t to know everything; it’s to be brutally honest about what you don’t know, and stay away from it until you’ve done the work. Over time, as you read, study companies, follow sectors, and make small, deliberate investments, that circle can expand.

Your mode of discipline should be:

Only make meaningful investments in areas you truly understand.

Size your positions smaller when you’re still learning.

Say “no” quickly to opportunities that sit outside your circle, no matter how exciting they sound.

Choose from ~50,000 publicly traded companies

The number of companies we can invest in today is truly staggering. There are almost 50,000 publicly listed businesses globally. Your job as an investor is to sift through that universe and find the ones that are either undervalued or, in some cases, clearly overvalued. With that much choice, most people don’t lose because there aren’t enough opportunities. They lose because they’re overwhelmed and don’t have a clear process for filtering what’s worth their attention.

The US is still the most liquid and widely followed market in the world, but keep in mind the world is a big place. Japan and Europe also offer deep, liquid markets. Beyond that, I’d strongly encourage you to look at selected emerging markets such as Saudi Arabia’s Tadawul (around 260 listed companies) and the UAE’s ADX and DFM (around 169 listed companies combined). In these markets, you’ll often find high-quality businesses that are followed by far fewer analysts and global investors than their US counterparts. That information gap is precisely where disciplined, long-term investors can sometimes find genuine mispricing.

It’s crucial to understand that emerging markets come with their own set of risks: currency swings, political and regulatory uncertainty, and, in some cases, weaker disclosure standards. The real question isn’t just “Where can I invest?” but “Which markets do I understand well enough to accept the risks and still sleep at night?”

A good starting point is this: pick two or three core markets. E.g. the US, one developed market (like Europe or Japan), and one emerging market (like Saudi Arabia or the UAE). Then build a focused watchlist of 20–30 companies across those markets that you gradually learn in depth. In a world where you’re spoilt for choice, having a clear map of which markets you invest in, and why, is one of the simplest edges you can give yourself.

Share

Understand enterprise value

Enterprise Value (EV) matters because it tells you what a business really costs to own, not just what the equity market says. Most beginners stop at market capitalisation:

Market cap = share price × number of shares.

That’s definitely useful, but it only tells you what you’re paying for - the equity slice of the business. In the real world, companies don’t exist as pure equity. They have:

• Cash (which reduces the “true” price you’re paying)

• Debt and other obligations (which increase the “true” price)

• Sometimes minority interests or preferred shares

Enterprise Value pulls all of that together.

Enterprise Value ≈ Market Cap + Total Debt – Cash and a few other items depending on how precise you want to be.

As an investor this is important because it reflects the price of the whole business, not just the shares. If you were buying 100% of a company in real life, you wouldn’t just pay the market cap and walk away. You’d also be taking on its debt and inheriting its cash pile. EV is the closest thing to a “takeover price” for the entire business.

It lets you compare companies with different capital structures. Two companies might look similar on a P/E basis, but one is loaded with debt while the other has a strong balance sheet and excess cash.

On P/E, they might both look “cheap”.

On EV/EBIT or EV/EBITDA, the highly leveraged one suddenly looks expensive once you factor in the debt. EV based multiples force you to acknowledge how a business is financed, not just how it earns.

It stops you from being fooled by “cheap-looking” stocks. A company can look cheap on market cap alone because its share price has fallen but if it’s drowning in debt, the enterprise you’re buying is still very expensive and very risky. Many value traps share this profile: low P/E, scary EV.

It makes more sense for comparing different sectors. In capital-intensive sectors (telecoms, utilities, airlines, energy, parts of emerging markets), debt is often a big part of the picture. EV-based multiples (EV/EBITDA, EV/EBIT, EV/FCF) are far more useful than simply looking at P/E or price-to-book in isolation.

It links naturally to how a real buyer thinks. Strategic buyers and private equity firms live in EV terms. They ask: “What’s the EV I’m paying?”, “What return am I getting on that EV given the cash flows (EBIT/EBITDA/FCF)?” Thinking in EV gets you thinking like a professional, rather than the trading mentality of “the stock looks cheap because it dropped 30%.”

That shift alone will put you ahead of most retail investors.

In the next instalment of this beginner’s guide I’ll be writing about the second most liquid and important market in the world; the global bond market.

Wishing you all the best,

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

Subscribe now

Watch the full episode here

Listen to the full podcast here

Follow Me On:
Substack
TikTok
X (Twitter)
Instagram
LinkedIn
YouTube

Themes: Artificial Intelligence, Data Centres, Leadership, GCC, UAE, Business, Investments

Subscribe now

Carl Agren is the co-founder of G42, a technology group based in Abu Dhabi that pioneers visionary artificial intelligence. He is the former CEO of Phoenix Group, a global digital infrastructure and data-centre company, and currently serves as the COO for Kevin O’Leary’s latest venture — Wonder Valley, the largest AI compute data-centre park on Earth.

In this interview, Carl shares a thought-provoking vision of the future — one that balances optimism about technology’s power to address humanity’s greatest challenges with a measured awareness of its potential to cause disruption and even dystopia if left unchecked.

Subscribe now

We talk about:

Carl’s journey as a leader building AI infrastructure and data centres
Building AI infrastructure in the GCC
What’s working — and what isn’t — with AI
How AI is fundamentally changing what it means to be a leader today
The impact of AI, AGI, superintelligence, and the singularity.

Share

Carl also discusses his next move with Wonder Valley, spearheaded by Kevin O’Leary’s investment group, which is envisioned as the world’s largest AI compute data-centre park, located in northern Alberta, Canada. The project aims to leverage the region’s abundant clean energy and cold climate to support large-scale artificial intelligence workloads with lower operational costs and reduced environmental impact.

The insights from this interview offer not only a deeper understanding of AI’s technological capabilities but also a reflection on the hierarchical and leadership implications of its growing dominance.

Follow Carl Agren:
Website
Instagram

Follow Me, Usman:
Substack
TikTok
X (Twitter)
Instagram
LinkedIn
YouTube

Themes: Artificial Intelligence, Data Centres, Leadership, GCC, UAE, Business, Investments

Subscribe now

Subscribe now

Carl Agren is the co-founder of G42, a technology group based in Abu Dhabi that pioneers visionary artificial intelligence. He is the former CEO of Phoenix Group, a global digital infrastructure and data-centre company, and currently serves as the COO for Kevin O’Leary’s latest venture — Wonder Valley, the largest AI compute data-centre park on Earth.

In this interview, Carl shares a thought-provoking vision of the future — one that balances optimism about technology’s power to address humanity’s greatest challenges with a measured awareness of its potential to cause disruption and even dystopia if left unchecked.

Subscribe now

We talk about:

Carl’s journey as a leader building AI infrastructure and data centres
Building AI infrastructure in the GCC
What’s working — and what isn’t — with AI
How AI is fundamentally changing what it means to be a leader today
The impact of AI, AGI, superintelligence, and the singularity.

Share

Carl also discusses his next move with Wonder Valley, spearheaded by Kevin O’Leary’s investment group, which is envisioned as the world’s largest AI compute data-centre park, located in northern Alberta, Canada. The project aims to leverage the region’s abundant clean energy and cold climate to support large-scale artificial intelligence workloads with lower operational costs and reduced environmental impact.

The insights from this interview offer not only a deeper understanding of AI’s technological capabilities but also a reflection on the hierarchical and leadership implications of its growing dominance.

Follow Carl Agren:
Website
Instagram

Follow Me, Usman:
Substack
TikTok
X (Twitter)
Instagram
LinkedIn
YouTube

Themes: Artificial Intelligence, Data Centres, Leadership, GCC, UAE, Business, Investments

That sounds dramatic, but you’re already seeing it:

• AI writing better first drafts than junior analysts.

• AI handling support tickets faster than entire customer service teams.

• AI building slide decks, financial models, marketing plans, even basic code.

If your day job is largely information in → process → information out, you are in direct competition with software that gets cheaper and more capable every year. Over the next decade, the market won’t pay a premium for things an AI can do in seconds.

Subscribe now

So if you can’t rely on your labour, what can you rely on?

For me, the conclusion is simple:

You need to own assets that earn while you sleep and appreciate over time.

That’s the only real hedge against a world where salaries are under pressure, career ladders are unstable, and “average” is automated away.

The Six Core Asset Classes

When we talk about “owning assets”, we’re essentially talking about six broad buckets:

1. Stocks (shares)

2. Bonds

3. Real estate (property) – residential & commercial

4. Commodities (gold, oil, etc.)

5. Crypto

6. Collectibles (art, watches, rare items, etc.)

All six can, in different ways, both produce income and appreciate in value.

But let’s be honest:
Two of them – real estate and collectibles – usually require large upfront capital, access to debt, and a pretty high tolerance for complexity and illiquidity. They’re great if you can play that game intelligently, but for most people starting out, they’re not the first move.

So in this post, I’m going to push you to focus hard on the remaining four, because almost anyone with discipline can build exposure to them over time:

• Stocks (shares)

• Bonds

• Commodities

• Crypto

Let me give a quick, no-nonsense lens on each.

Stocks: Owning Pieces of Real Businesses

When you buy shares, you’re not buying squiggly lines on a chart, you’re buying a slice of a real business:

• A company that sells products or services.

• Has customers, employees, assets, debts.

• Generates (or should generate) free cash flow.

If AI is going to eat jobs, it won’t eat all jobs equally. The biggest winners will be well-run companies that can use AI to reduce costs, scale faster, and capture more profit.

Owning stocks is essentially saying:

“Instead of competing with AI as labour, I’d like to own the businesses that are deploying AI.”

In future posts, I’ll break down:

• How to read and build financial models in a simple and straight forward way.

• How to think about valuation and forecasting (P/E, earnings yield, DCF, etc.).

• How to separate hype from genuine business quality (understanding business fundamentals).

  Subscribe now

Bonds: Getting Paid To Lend Money

Bonds are simpler than people make them:

• You lend money to a government or a company.

• They agree to pay you interest and return your principal at maturity.

In a world where job income is uncertain, fixed income can be a stabiliser in your portfolio:

• It won’t make you rich quickly.

• But it can provide predictable cash flows and reduce volatility.

I’ll cover how to think about:

• Government vs corporate bonds.

• Credit risk and default risk.

• How bonds behave when interest rates move.

Commodities: Owning the Raw Inputs

Commodities are things the real economy needs:

• Energy (oil, gas).

• Metals (gold, copper).

• Agricultural products.

They usually don’t generate cash flow by themselves (unless accessed via producers or structured products), but they can:

• Hedge inflation.

• Provide diversification when financial assets are volatile.

• Sometimes massively reprice when supply/demand shocks hit.

The key is understanding why you’d own them and how (direct exposure [futures contracts] vs companies vs ETFs).

Crypto: Asymmetric Bets in a New System

Crypto is the most controversial of the list, but it’s here for a reason.

Whatever your view, crypto represents:

• A parallel financial system being built in real time.

• A set of assets with high volatility and asymmetric payoff profiles.

You do not go all-in. But a small, intelligently structured allocation can:

• Expose you to new financial rails and technologies.

• Potentially provide outsized upside if certain narratives (digital gold, smart contract platforms, tokenisation, government / institutional adoption etc.) play out.

  Share

I’ll talk about:

• How I think about position sizing in something this volatile.

• The difference between speculation and thesis-driven investing.

• Basic risk controls so you don’t blow yourself up.

“But What About Owning a Business?”

A fair question.

If assets are so important, why not just own a business and capture all that upside yourself?

Here’s the uncomfortable truth from the lens of someone who runs a management consulting firm and spends their time looking under the hood of other companies:

Most small businesses are not profitable. And a scary number of them are barely surviving.

Why?

• Risk: Customer concentration, key-person dependency, regulatory shocks.

• Competition: Intense global and local competition, low barriers to entry, price wars.

• Capital: You often need upfront investment long before the model is proven.

• Time to cash flow: It can take years before a business consistently pays you instead of draining you.

• Brand and distribution: Great product, no attention = no revenue.

From what I’ve seen, easily 80%+ of small businesses are struggling i.e. not compounding. Owners are stressed, overworked, and under-paid relative to the risk they’re taking.

That doesn’t mean “don’t start a business”. It means:

• If you do, treat it as a high-risk, high-effort venture, not your only plan.

• And regardless of whether you’re a founder, employee, or freelancer, you should still be building an asset base outside your primary income.

The beauty of stocks, bonds, commodities and even crypto is:

• You can access them with relatively small ticket sizes.

• You can diversify across sectors, geographies, and themes.

• You are not responsible for payroll, rent, marketing, operations, management etc.

You’re not giving up on entrepreneurship. You’re just not betting your entire future on one fragile choice.

Why You Should Listen to Me

A bit of context.

I’ve spent 15+ years in finance and investment management, and I now run a management consulting business.

That gives me a fairly unique mix of experience which nobody else has:

• On the markets side: understanding companies through their financial statements, valuations, cash flows, and capital structure. As well as understanding how the financial markets tend to work by asset class and risk / return scenarios.

• On the business side: seeing how companies actually make or lose money in the real world; pricing, operations, strategy, leadership, and all the unpredictable human related issues.

In this Substack I’m going to combine those two lenses.

Over the coming posts, my goal is to help you:

• Understand how AI is reshaping and disrupting careers, companies, and capital.

• Learn how to analyse businesses like an investor, not a speculator. Sorry, there’s no day trading here.

• Build a personal asset base, primarily via stocks, bonds, commodities, and crypto, that can realistically change your trajectory over the next 5–20+ years.

If you’re worried about being made obsolete by AI, then you should consider subscribing for the free occasional posts or become a paid member for deeper dives into how I research and analyse asset classes alongside build portfolios for the best risk/return scenarios.

Subscribe now

Own more of the system and rely less on your job.

This publication is my attempt to show you, step by step, how to do that.

Hope to see you soon.

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

The post below summarises what most tell me off camera and behind the scenes.

Our purpose and existence are currently being indirectly threatened by AI and machines. We’re being told that we shouldn’t worry as we can simply learn to use AI, which will help us enhance our productivity in our jobs and businesses. However, that couldn’t be further from the truth.

AI is advancing more quickly than we can imagine. It is learning at an almost exponential rate. In fact, according to the IMF, approximately 40% of jobs globally are exposed to AI, and ~60% in advanced economies. About half of exposed roles benefit, and the other half face lower demand and wages. Estimates from Goldman Sachs suggest up to 300 million full-time job equivalents could be automated worldwide. And all of this has primarily taken place since 2022, when OpenAI released ChatGPT to the world.

Across the OECD, about 28% of jobs sit in occupations at high risk of automation, and roles such as clerical/administrative work are the most exposed. And because these roles are disproportionately held by women, the impact is gendered.

What AI job losses?

Currently, most of the white-collar job cuts in the US and Europe are still being driven by old-school reasons: cost-cutting, slower demand, higher interest rates, and major corporate restructurings. In the US alone, employers have announced roughly 1.1 million job cuts in 2025 so far, and the main reason cited is cost reduction, with AI coming in as the second-most cited factor, not the first.

When you zoom in on AI specifically, the numbers are big enough to matter, but still a minority of total cuts. Challenger, Gray & Christmas report that in October 2025, companies explicitly blamed AI for 31,039 layoffs in a single month, and 48,414 cuts so far this year in the US. That’s roughly a low single-digit percentage of all announced job losses.

If you widen the lens from “AI” to “automation and tech updates,” the picture gets a bit larger. Challenger data for 2025 shows 20,219 job cuts attributed to “technological updates, including automation and possibly AI,” on top of the 17,000+ job cuts explicitly tied to AI earlier in the year. Put simply: tens of thousands of roles are being directly rationalised by automation and AI, within a broader pool of more than a million cuts driven by cost, restructuring, and weak demand.

You can also point to concrete corporate moves that are explicitly about swapping humans for software. IBM’s CEO has said around 7,800 back-office jobs at IBM could be replaced by AI in the next few years, and the company has already slowed hiring in those functions. BT in the UK plans to cut up to 55,000 jobs by 2030 and expects about 10,000 of those roles, mainly in customer service and network operations, to be replaced by AI and digital tools. Klarna’s AI assistant now does the equivalent work of 700 full-time customer service agents, after the company restructured its service workforce around the bot. Amazon just announced 14,000 corporate job cuts while simultaneously ramping up multi-billion-dollar investments in AI infrastructure.

In Europe, the pattern is similar: economic tightening + AI. A recent survey by the UK’s CIPD found that one in six UK employers expect to cut jobs in the next year specifically because of AI, with the biggest risk falling on clerical, junior managerial, and administrative roles. European policy analysis noted that rising borrowing costs and the rapid roll-out of AI have “converged to trigger a wave of job losses” that is reshaping white-collar work across the continent.

Forward-looking surveys show where this is heading. The World Economic Forum’s Future of Jobs 2025 report finds that 40% of employers globally expect to reduce their workforce where AI can automate tasks, even as technology simultaneously creates new roles. Some experts now openly talk about the possibility that up to half of entry-level white-collar jobs could be automated away in the next five years, particularly in law, finance, consulting, and tech.

So the honest answer to “are these white-collar job losses happening because of AI?” is: partly. The majority of current layoffs in the US and Europe are still being justified as cost-cutting, restructuring, and macro-economics. But improvements in digital transformation, most of which now involve AI, also hide the true cause behind many job cuts. AI is no longer a hypothetical threat; it is already responsible for tens of thousands of documented job cuts, and it quietly underpins many cost-cutting decisions that don’t explicitly mention it.

The sharper story is this: AI is simultaneously removing some roles outright and closing the door on new hiring at the bottom, which is exactly what entry-level white-collar workers should be worried about.

AI disguised as “upgraded tech”

AI is being rolled out under a very friendly label: “upgraded tech.” On paper it’s just the next version of the tools we already use: a smarter CRM, a better helpdesk, a “Copilot” sitting next to you in Outlook or Excel. The marketing is deliberately soft. You’re not being told that your role is on the line; you’re being told you’re getting a productivity boost, a digital assistant, an upgrade. The word “automation” appears, but usually buried under a lot of talk about “freeing people up for higher-value work.”

But underneath the branding, AI is not just an upgrade, it’s a substitution. When a company introduces AI that drafts emails, summarises meetings, handles customer chats end-to-end, reconciles invoices, or updates systems automatically, it is quietly doing two things at once: helping existing staff for now, and testing how many of those staff it can live without later.

The first phase is always framed as experimentation and augmentation. The second phase shows up as “restructuring,” “rightsizing,” and “not backfilling” junior roles. The technology didn’t change between those two phases, but the story did.

That’s what I mean when I say AI is disguised as upgraded tech. On the surface it looks like a nicer version of yesterday’s software. In reality, it lets management ask a more brutal question: if this system can do 30–50% of the work of an average employee, why am I still paying for 100% of that salary?

You’ll rarely hear it expressed that bluntly in public, but you see it in the hiring freezes at the bottom of the ladder, the quiet disappearance of entry-level and graduate jobs, and the increasing pressure on the people who remain. The interface looks like an upgrade however the balance of power underneath, not so much.

Subscribe now

What about artificial general intelligence (AGI)?

AGI is what shows up when people stop talking about tools and start talking about replacements. Narrow AI handles specific tasks i.e., summarising documents, answering support tickets, generating code etc. AGI is the idea of a system that can match or exceed humans across most cognitive domains: reasoning, planning, learning, problem-solving not just autocomplete your emails.

Depending on who you listen to, this is either decades away, already emerging, or quietly here in limited form. Surveys of AI researchers typically cluster median AGI timelines somewhere between the 2030s and 2050s, but with huge disagreement, which is another way of saying nobody actually knows, but a lot of serious people think it’s plausible in your working lifetime.

From my discussions with numerous AI and tech founders/CEOs, a form of AGI is already here in some capacities but not others. As an example, my consulting firm has a partnership with a small Italian AI start-up called Kapto. On the face of it, Kapto looks like a standard intelligent document processing system, however on deeper inspection the software is sophisticated enough that it can ingest documents, analyse, reason, and decide what outcome the data should conclude.

So for the insurance industry, claims handling and underwriting, the system can work the life cycle of a claim from inception to approval or decline. And it’s already being rolled out in insurance companies across Europe, where the need for large claims and underwriting teams is now unnecessary. These are entire departments of people being made obsolete.

The moment AI can reason with judgement based on finite data i.e., being intuitive enough to reach conclusions that most intelligent and domain-expert humans are expected to reach, is where AGI starts to poke its head. This may not be an “official launch,” however, it is an iterative process emerging from smaller applications.

What matters for your career isn’t the exact date, it’s the direction of travel. We already have models that can pass bar exams, write production-grade code, generate photorealistic video, and autonomously plan and execute multi-step tasks through agents. Each new generation absorbs more of what used to be uniquely human: pattern recognition, writing, strategy, even basic creativity.

If you extrapolate that curve even modestly, you don’t need full-blown sci-fi AGI to cause serious disruption. A world where AI systems are ‘good enough’ to handle 70–80% of professional knowledge work is already existential.

The CEOs I speak to aren’t lying awake at night wondering about the philosophical definition of AGI; they’re asking, very practically: how far can we push this before we hit limits or regulators?

There’s also an uncomfortable asymmetry built into the AGI conversation. If AGI is further away than the hype suggests, we’ve “only” got to deal with decades of automation, job polarisation, and constant re-skilling.

If the optimists are right and something AGI-like arrives sooner, we’re talking about a system that could, in principle, learn any digital skill faster, cheaper, and at larger scale than humans, including the skills we currently treat as our safety net. That doesn’t mean a Hollywood-style apocalypse, but it does mean the old promise “just keep up-skilling and you’ll be fine” starts to sound weaker.

The rational stance isn’t panic, but honesty: we are deliberately building systems that may outgrow us, and pretending that’s just another “tech upgrade” is a luxury we don’t have.

Ok, so are we all doomed?

Like anything, it depends. If you’re reading this in 2030 or beyond and still thinking about what to do, then I really feel bad for you. However, if you start making inroads now and planning for your future, there is still a chance to get ahead and “beat AI” before obsolescence.

I am sure there are going to be many areas of employment and business created due to the rise of AI and AGI, however it’ll only lead to a temporary gap filled by products or services until superintelligence arrives.

Superintelligence is a hypothetical agent that surpasses human intelligence in virtually all cognitive domains, including creativity, problem-solving, and emotional intelligence. Unlike current AI, which is specialised for single tasks, superintelligence would be a general-purpose intellect far beyond what humans can achieve.

Once systems move from “very smart tools” to something closer to superintelligence, most of the new roles AI created on the way up will also be on the chopping block. Prompt engineers, AI product managers, even today’s “AI safety” specialists are, in a sense, transitional jobs. They exist to bridge the gap between dumb software and something that can largely run itself.

That doesn’t mean there’s zero opportunity, there will be fortunes made in that transition, but it does mean you shouldn’t build your entire life plan on being a slightly more efficient cog in an AI-driven machine. Your edge can’t just be “I know how to use the tools”, the tools will get better faster than you can.

What you can do now

If you’ve read this far, you already understand the gravity of what’s coming. AI isn’t just changing work, it’s rewiring who wins and who gets left behind. The people who thrive in the next decade won’t be the ones with the longest CVs. They’ll be the ones who understand capital, ownership, investing, business models, and how to build leverage in a world where machines do most of the execution.

And those are skills nobody teaches you in school, university, or in your job.

That’s why I built this Substack. If you want to move from an “employee mindset” to an “owner mindset,” I’m building a full, practical curriculum here: how to invest intelligently, how to analyse companies, how to build real wealth through ownership, and how to understand business models so you can spot opportunity instead of fearing the future.

Paid subscribers will get step-by-step guides, investing frameworks, financial models of actual investment opportunities, breakdowns of real companies, and actionable lessons you can immediately apply to build financial resilience in an AI-dominated world.

Subscribe now

Hope to see you soon.

Usman

Note: I never give financial advice, and I’ll never DM you asking for payment. All activity happens here on Substack.

P.S. If you use Gmail and have 10 seconds to do me a small favour (to help convince Gmail to keep this newsletter out of the Promotions folder)

1. Reply to this email with a quick “Hi!”

2. Move the email to your Primary folder

1. AI, Finance & Strategy Desk Notes

Short, sharp insights on AI breakthroughs, market developments, business models, and future-of-work trends.

✅ My day-to-day take on how AI is reshaping industries, jobs, wealth, and opportunity.

2. The Owner’s Playbook (Educational Investing Portfolio)

Illustrative breakdowns of companies, sectors, and investment opportunities built to teach you how to think like a value investor, not follow tips.

✅ One-way feed: frameworks, models, valuation walk-throughs, and clear reasoning behind every example.

3. Education, Frameworks & Leverage

Deep-dive lessons on capital, financial modelling like a hedge fund analyst, ownership, business strategy, and financial intelligence.

✅ Learn how top operators, investors, and CEOs think — and how to build leverage in a world dominated by AI.

4. Community Access

Ask me questions, request breakdowns, and join discussions with others building wealth in the AI era.

✅ Engage in daily learning, share insights, and grow with a community that thinks long-term.

***How It Works***

✅ Paid Members: full access to The Owner’s Playbook, all educational content, deep-dive guides, step-by-step frameworks, and priority Q&As.

ATTENTION: All content here is for educational and informational purposes only never financial or investment advice.