How the Rotation Into "Safe" Stocks Gone Too Far?

Last year, investors were obsessed with AI. The Magnificent Seven rose 23% in 2025 alone, adding $4 trillion in market cap. 

But the vibe has shifted, and the tech sector has become the ugly duckling. Software stocks have undergone the largest nonrecessionary 12-month drawdown in over 30 years, shedding more than $1.3 trillion in market value. 

Investors have piled into traditionally “boring” sectors. Consumer staples are up nearly 14%, materials are up 18%, and energy is up 22%.

Amid unprecedented AI spending, these stocks now look more appealing for their conservatism. 

But now the market has rotated so fast into safe stocks that even old stalwarts start to look expensive and, well, less safe.

Investors are now  paying a premium now for “AI immunity,” companies that sell toothpaste and detergent and have razor-thin margins. Meanwhile, SaaS companies that have 80% gross margins, recurring revenue models, real pricing power, and network effects that create moats are trading at a huge discount. 

There are real opportunities in these fallen angels. Let’s talk numbers. Generally speaking, the valuation of a company should be roughly the asset value plus the present value of the growth opportunity. The key word there is growth. Growth has a tendency to solve all problems. If you look at the price/earnings-to-growth (PEG) ratio for the “safe” sectors versus tech, you’re paying three to five times more per unit of growth for toilet paper than you are for enterprise software.

When wisdom becomes conventional, it’s no longer wise. There’s this trope that people will always need toothpaste and shampoo and that therefore these companies are recession proof. That is total BS. 

You know what is more recession proof? Enterprise software that runs mission critical operations in a company. When a recession hits, consumers stop buying lattes, companies stop buying new office chairs, but companies don’t stop using Salesforce to manage their customer pipeline.

The price/earnings-to-growth (PEG) ratio is a stock’s price-to-earnings ratio divided by its expected earnings growth rate. The PEG ratio is a more nuanced way of looking at a stock’s value than PE ratio because it takes growth into account. 

For example:

Company A: P/E = 20, expected EPS growth = 20% → PEG = 1.

Company B: P/E = 20, growth = 40% → PEG = 0.5.

Despite having the same P/E, Company B is cheaper relative to its growth; you’re paying the same $20 per $1 of earnings, but you’re getting twice the growth.

The most underhyped, overpunished software category is cybersecurity. Here’s why: 

AI is making cybersecurity threats worse, and companies aren’t just going to vibe code their own protection. In the same way that AI can make a bad coder pretty good, AI can turn mediocre cybercriminals into very effective cybercriminals, and make the good ones even better. 

According to Help Net Security, the fastest 25% of intrusions now steal data in 1.2 hours — down from 4.8 hours just a year before.

AI can write and launch a sophisticated phishing campaign in just five minutes — a task that used to take a team of humans 16 hours. Since the launch of ChatGPT, phishing attack volume has increased more than 4,000%.

Sidenote: Whoever is phishing Prof G Media employees, please continue using the following signoff; it’s super effective:

Two-thirds of organizations globally are planning to increase their investment in cyber risk prevention over the next 12 months to address this risk.

The second factor: The Trump administration is preparing to include private cyber companies in offensive cybersecurity for the first time. This would open up companies to billions in new government contracts.

That brings me to the most obvious choice in the category: Microsoft. The most underappreciated story about the stepfather of OpenAI is that it is also a cybersecurity company. 

In fact, it’s the largest cybersecurity vendor in the world. It made an estimated $37 billion from its cybersecurity products in 2025 — at least three to four times larger than any pure-play competitor. 

Also: More than 85% of the Fortune 500 uses Microsoft 365 software, it has the second-largest cloud business in the world, it’s the third-largest gaming company in the world, it has the world’s largest professional network (LinkedIn), it owns 27% of OpenAI and is entitled to 20% of its revenues through 2032. 

It’s currently down 18% year to date and trading at 25 times earnings, the lowest it’s been since 2022.

Wealth taxes are back in the spotlight. 

In the Netherlands, lawmakers just approved a plan to tax unrealized gains on assets like stocks, bonds, and crypto. Back in September, a French economist proposed a 2% tax on French residents with over €100 million in assets. 

In California, a proposed 2026 ballot measure would impose a 5% one-time tax on individuals worth more than $1 billion. That proposal in particular has sparked a lot of debate. Kalshi briefly priced the odds of the measure passing at 45% after Bernie Sanders headlined a support rally last week; since then, they’ve dropped back to 34%. 

Wealth inequality in the U.S. has never been higher. The top 19 households control roughly 2% of all household wealth in America — up almost 20 times in the past 40 years. The bottom 50% of Americans control just 3% of household wealth, which means that 19 households have roughly two-thirds as much money as 65 million households.

Taxing the rich is politically popular; 60% of likely California voters back a one-time wealth tax, even as a majority of those same respondents admit the move would spark a business exodus and cost local jobs. 

However, it’s unpopular among the voters who have the most influence: the ultrawealthy.

The richest Americans can pay to influence elected representatives, and if that doesn’t work, they can hire tax experts who take advantage of loopholes in America’s overly complex tax laws. 

The problem with wealth taxes is that billionaires will immediately hire the best tax attorneys and argue their assets are worth 40% of what the government says they are. Valuations would be an absolute nightmare. How do you value a Picasso? Is it worth $3 million or $30 million? How would they value my stake in Prof G Media? 

You would literally need 10 times the budget of the IRS to administer a wealth tax, and most of the money would go to paying lawyers to argue about yacht valuations. 

In addition, wealth taxes don’t work. In 1995, there were 12 developed countries with wealth taxes — now, only three remain. The uberwealthy are the most mobile people in the world, and they can just leave.

There’s an important difference between being right and effective. The super rich are not paying their fair share. That’s true, and we’ve got to do something.

Here’s an idea. Get rid of the carried interest loophole and get rid of the lower capital gains tax. There’s no reason I should be paying 21% when I sell my stocks while you pay 37% when you make money. There are much more pragmatic and enforceable means of raising taxes on wealthy people.

Monetize Your Expertise, On Your Own Time.

Receive $125 for 25 minutes of your time, plus real-time access to industry-leading research.

By participating in Qualitate’s market research surveys, access insights like this and more:

If you have visibility and responsibility into which software solutions your organization is using, we want your perspective.

Receive $125 after a brief 25 minute survey, plus real-time access to anonymized industry-leading research trusted by the world's largest enterprises and investment firms. Participate in the first 48 hours and receive an exclusive bonus.

Contribute your expertise. Access the results. Stay ahead.

Join the study here.

AI is quickly becoming one of the defining political issues of the decade. What once looked like a breakthrough technology with broad upside is now drawing scrutiny over its real-world costs, from energy demand and infrastructure strain to concerns about jobs and economic disruption. 

Data centers in particular are becoming political flashpoints. Construction workers benefit in the short term when a data center comes to town, but these facilities generate only about 100 jobs permanently — that’s about a third of the number of people that work at an average Walmart. 

The real problem is that they strain the power grid, making electricity a lot more expensive. The Stargate Project’s main data center in Texas sits on a plot of farmland bigger than 1,000 football fields and will use as much electricity as all the homes in Philadelphia.

Protests are stacking up. Grassroots organizers have blocked $64 billion worth of data center projects in just two years. There are now at least 142 activist groups across 24 states organizing against data center construction, six states have bills to pause it entirely, and residents in Wisconsin are trying to recall their mayor for allowing an OpenAI data center to get built.

Millions of households have seen their electricity rates go up 67% because there’s some empty building down the road full of computers and chips, not employing anybody and sucking all the energy off the grid. If AI isn’t going to make the economic lives of ordinary people easier, it at least can’t raise their prices. Simple legislation would be to start taxing data centers, at least to the incremental cost of the energy and the price increases they’re causing.

This is rapidly becoming the biggest issue in American politics right now. Every elected official is trying to decide their stance on AI, and my prediction is that you’re going to see a lot more politicians coming out against it.

Wall Street is underestimating the potential impact of this. What if most Americans decide they don’t want a data center in their backyard? If that happens, what does that do to the future cash flows of a Google or an OpenAI? Does that hurt future revenue growth? I would argue yes.

In a 6-3 decision announced on Friday, the Supreme Court ruled that Trump exceeded his authority by imposing sweeping tariffs under a 1977 law called the International Emergency Economic Powers Act. 

The ruling invalidates his “reciprocal" country-by-country tariffs and the fentanyl-related 25% tariffs on Canada, China, and Mexico, but leaves intact tariffs imposed under other laws, like those on steel and aluminum. 

Trump called the decision a “disgrace” and said he’ll impose a new 10% global tariff using a different law that limits tariffs to 150 days, unless extended by Congress. The following day, Trump said he wants a global tariff of 15%.

The questions we’ll be watching as this plays out:

Writing will produce forgetfulness in the minds of those who learn to use it, because they will not practice their memory. … You have invented an elixir not of memory, but of reminding; and you offer your pupils the appearance of wisdom, not true wisdom.

— Socrates

Socrates got it wrong. So did the catastrophizers who warned us about radio, calculators, TV, and even search engines. Google didn’t make us that dumb, so the logic follows that AI won’t make us dumb either. This is false comfort. The conviction that history always repeats itself is getting in the way of our realizing that this time is actually different. 

Last month, Dr. Jared Cooney Horvath, a neuroscientist and educator, told Congress that for the first time since the 1800s, Gen Z is not as smart as previous generations were at their age. Specifically, Horvath said, Gen Z underperforms “on basically every cognitive measure” ranging from executive function, and literacy, to general IQ, memory, and attention skills.

We can’t fully blame AI. After all, it’s only been with us for the last four years. This decline began with the internet and got worse with smartphones. What happened? Instant gratification and constant multitasking became the new normal, making it nearly impossible for people to concentrate for longer than 47 seconds. Getting smarter requires being able to pay attention, and we’re increasingly distracted. 

It’s not just Gen Z. Adults have gotten less intelligent, too. AI threatens to make this worse.

Study after study has shown that when we don’t use skills, we lose them. AI isn’t just a calculator or a search engine — it’s a critical thinking machine that can do pretty much anything we ask it. Consistently outsourcing the process of thinking … will change our brains and change our society — for the worse. 

Psychologists described the consequences: “AI assistants are designed to mimic cognitive skills. … For this reason, consistent and repeated engagement with an AI assistant is likely to lead to greater decrements in skill.”

I'm not a Luddite — I believe that AI is a remarkable and inevitable technology, and everyone needs to understand how to use it. It’s also just a really convenient time-saver. But what’s not being acknowledged is that while AI might technically be free, it has a hidden price. If you want or need to outsource your work to AI, you should at least understand what you’re paying. 

In the full article on Substack, I break down the cost using research-based studies on how AI impacts writing, learning, and critical thinking. (Hint: It’s not good.)

We are going to bomb Iran. We have 13 warships in the Middle East, one carrier strike group already there, and the USS Gerald R. Ford, the world’'s largest aircraft carrier, is currently en route. Diplomatic talks are ongoing, but described as very far apart.

In this special Lunar New Year episode of China Decode, Alice Han unpacks the meaning — and the mythology — behind the Year of the Fire Horse. Astrology, politics, and economics are intersecting at a pivotal moment. What does this year reveal about where China may be headed next?

INVESTMENT DISCLAIMER:

The opinions in this newsletter are for informational purposes only. They are not investment advice.