Start - Björn Jeffery

Any other answer would have been a shock

June 2, 2026

This analysis was first published in SvD Näringsliv, in Swedish, on June 2nd, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

Is social media harmful to young people? Sweden’s investigators have delivered a much-anticipated answer. Anything other than the conclusion they reached would have been a shock – given how the assignment was framed. But the most important question remains unanswered.

Ask a question, get an answer. The government version of that might be: commission an inquiry, get an inquiry.

When Minister for Social Affairs Jakob Forssmed appointed a special investigator last fall to figure out how an age limit on social media could be implemented in Sweden, the assignment was crystal clear:

“Harmful content and addictive algorithms must be pushed back to strengthen the protection of children’s health, safety, and wellbeing. We are now investigating how a strict age limit on social media can be designed in Sweden.”

The question of whether social media is actually dangerous was set aside in favor of how an age limit could be implemented. But that we still got no answers on in the interim report released Tuesday.

In an op-ed in SvD, the investigators outline their findings. The legal conditions for a Swedish age limit on social media have been reviewed, and the conclusion is that it’s feasible. Legally speaking. A limit of 15 years is proposed. Snapchat, Instagram, Facebook, TikTok, and YouTube are among the platforms that would be covered.

Beyond that, the investigators were also asked to consider whether this is something that should be introduced at all. That’s a considerably bigger question. And, unsurprisingly, they concluded that the proposal both can and should be implemented.

Any other answer – given how the question was framed from the start – would have been a shock.

Public opinion on this issue is also entirely one-sided right now. As the investigators themselves note, similar proposals are already finalized in several European countries, and there are long lists of ongoing inquiries of the same kind elsewhere. Out in front is Australia, which in December 2025 became the first country in the world to introduce a ban on social media for children.

Bucking that trend would have required enormous courage. And that’s exactly why it’s particularly interesting to look at the reasoning for why this is such a good idea – beyond the fact that many other countries seem to think the same. It’s also worth examining how things have played out in Australia so far, since they’re the only country that has actually implemented such an age limit.

Let’s start with the land down under. In March, a report from the Australian government looked at exactly this. It found that around 5 million social media accounts have been removed. That sounds like a lot. But the report acknowledges that the figure doesn’t correspond to the number of young people who have lost access to social media, since the same person can have multiple accounts. And more importantly – seven in ten children still had social media accounts despite the law.

One of the goals of the age limit was to reduce bullying and harmful behavior through these channels. The Australian report shows, however, that the number of reports has not decreased since the ban was introduced. It’s still early – the law is barely six months old – but so far, the problems appear to persist.

In their op-ed, the investigators write that “the research on children’s wellbeing and their use of social media [is] clear.”

Not all researchers would agree with that claim. Developmental psychologist Candice Odgers at the University of California, Irvine, is one of them.

She has studied young people’s use of smartphones since 2008 and argues that research – including a meta-study examining 226 studies over 12 years – cannot demonstrate any connection between social media use and wellbeing at all. Research from Amy Orben and Andrew Przybylski at Oxford University shows similar results. We can see that many young people are struggling, suffering from anxiety and depression. But that social media is to blame is far less clear.

There is, of course, research pointing in the opposite direction too. And that seems to be what Minister Forssmed and the investigators have chosen to focus on. That’s a valid choice – but to say the research is “clear” is a significant oversimplification.

The same thing happened when Sweden’s Public Health Agency issued screen-time recommendations. What was proposed had weak backing in the very research evidence they themselves provided.

One of the biggest questions – even with an age limit in place – is how this would work in practice. How do you even verify users’ ages accurately? The investigators also raise their own concerns about the need to preserve anonymity on social media. But these questions are once again left unanswered. “The inquiry will in its next phase continue to follow that work closely,” the report says, referring to the EU’s work on digital age verification.

Protecting young people is a noble mission that most people can get behind. And few today believe that unregulated tech companies would prioritize young people’s wellbeing. The work is important and affects almost everyone in Sweden.

The real question is how we as a society actually get this right. How we improve young people’s wellbeing without creating other, unintended problems. Or for that matter – whether we can even identify an approach that genuinely works.

This inquiry gave us – once again, and despite good intentions – very few answers on how to do that.

Writers vs AI — the battle is already over

May 29, 2026

by Björn Jeffery

Leave a comment

SvD Näringsliv

Idealism meets pragmatism as the Swedish Writers’ Union gathers its members for a meeting about the AI future. The mood is set to be tense — but the central conflict has already been partly decided.

It’s not surprising that the tech world is so disliked across large parts of society. Every new technology shift comes with promises of riches and a brighter future — last time it was social media, this time it’s AI.

The riches, however, mostly seem to stay in Silicon Valley. And the broader benefits leave plenty of question marks too. It’s a long way from Google’s Nobel-winning Alphafold project — the AI system that can compute protein structures — to using enormous amounts of electricity to generate strange memes and AI slop.

It’s understandable, then, that Swedish authors aren’t cheering when the conversation turns to their works being included in a new Swedish AI model. The goal is to produce a language model that understands Swedish deeply and that isn’t dependent on American tech companies.

The technology has underdelivered before. When the Swedish Writers’ Union gathers on Saturday to discuss the issue, the mood is set to be sour.

At the same time, there seems to be a misunderstanding both about how the technology and the market function, how they could function — and what’s actually happening right now. Whether you like it or not.

One of the project’s critics, the poet Olivia Bergdahl, tells Aftonbladet: “The way these machines work, it’s the books that get used, but really it becomes more a question of what price tag you put on the language an author has developed — and that’s impossible.”

That may be true. But the price tag on the existing works that have found their way into today’s AI models is already set. It’s exactly zero kronor. So we’re not talking about a situation where authors’ books might end up in AI models — they’re already there. Along with a long list of other copyrighted material that has also ended up there, without permission.

Last autumn, AI company Anthropic reached a settlement in which it agreed to pay 1.5 billion dollars for having used books without permission to train its AI models. The examples are many, and they exist in adjacent industries too. The New York Times, for instance, has sued OpenAI for taking its journalism in a case still working its way through the American legal system.

The conversations the Swedish Writers’ Union is trying to have about a new Swedish language model can be seen in this light. The initiative looks less like a way to embrace AI technology and more like an attempt to step a couple of paces back from where we are today. The aim is to be able to regulate compensation and the use of books in these models under far more controlled terms than what’s happening now.

If this sounds like a kind of hostage situation, there’s a reason for that. American and Chinese companies are currently taking Swedish and foreign literature and training their AI models on it. No one — neither the authors themselves nor the Writers’ Union — has any say in that fact. That obviously doesn’t mean you have to welcome this future with open arms and simply accept it. But to realistically argue for staying entirely outside it, you have to think about how that would actually work in practice. It might mean suing the AI companies in question. But even that is something you’d most likely want to do as a collective rather than as an individual author.

The Swedish Writers’ Union has, thanklessly, ended up taking the blame for this messy AI era. What they’re discussing is participating in a Swedish research project where a different kind of control around compensation and copyright could be built.

If members decide — now or in the future — that they want to fully withdraw from all AI involvement, there’s the option of finding solutions for that too. And if large-scale copyright infringement lawsuits also arrive in Sweden, it might be wise to have a partner who understands how the technology works from the inside.

That holds — unfortunately, you might say — even for those who want nothing to do with it.

This analysis was first published in SvD Näringsliv, in Swedish, on May 29th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

A brutal wake-up call for Kinnevik’s portfolio

May 25, 2026

by Björn Jeffery

Leave a comment

SvD Näringsliv

When Helena Saxon steps in as CEO of Kinnevik, the company’s tech strategy is buried once and for all. A new reality awaits — and some are in for a brutal wake-up call.

When Georgi Ganev, then-CEO of Kinnevik, summarized the year 2021, he wrote the following as the opening line of the quarterly report:

“During 2021, we significantly reweighted our portfolio composition toward younger, private growth companies.”

The mandate for incoming CEO Helena Saxon appears to be the exact opposite. And if you’re a company sitting in Kinnevik’s current portfolio, the future suddenly looks anything but secure.

Reading chair Cristina Stenbeck’s statement in the press release, the new direction is crystal clear:

“We will invest primarily in profitable, cash-flow-generating growth companies where growth capital is scarce and where we can build successful and sustainable businesses across generations.”

If we allow ourselves to paraphrase the somewhat dry quote, we can sum it up like this instead: things at Kinnevik are about to get stable, predictable, and a little boring.

And it’s hard to read the statement as anything other than a repudiation of Kinnevik’s strategy in recent years.

Let’s recall what happened: in recent years, Kinnevik has invested billions in precisely the kind of unprofitable companies that need enormous amounts of growth capital, that haven’t generated a single krona in cash flow, and whose sustainability — and in some cases even their temporary success — has been heavily questioned.

This is a new and inverted strategy.

The market clearly agrees with the verdict. Over the past five years, the stock has fallen by more than 80 percent. Helena Saxon’s previous employer, Investor, has doubled its value over the same period.

Other investment companies on the Stockholm stock exchange haven’t performed as well as Investor, but they haven’t done as poorly as Kinnevik either. With one exception — VNV Global — whose focus has been very similar to Kinnevik’s, with bets on unlisted tech growth companies. The fund Tin Ny Teknik, which pursued a similar strategy, has also lost 57 percent of its value over five years.

If we want to be generous to Kinnevik, we can note that it wasn’t only the execution of the strategy that was the problem — it was the timing too. They were too late to catch the big winners — Spotify, Klarna, King — and they spent too much money on inflated valuations of companies that never lived up to those predecessors.

When it comes to the new generation of AI companies, Kinnevik has put money into Tandem Health — but missed the biggest Swedish winners so far, Lovable and Legora. VNV and Tin Ny Teknik are also absent there.

The big question becomes what happens to the existing portfolio. The list of candidates that can start generating large cash flows any time soon is short. Stenbeck herself writes that they “will be very disciplined when it comes to follow-on investments” and that they will “work with our portfolio companies to maximize each company’s potential and drive positive development.”

But positive development for whom? For Kinnevik, it could just as easily mean selling off these unprofitable holdings. A couple of fintech names were already sold last year, in a deal that short-seller firm Ningi Research criticized. These companies depend heavily on follow-on investments to keep operating. And having a major shareholder that doesn’t want to participate sends a negative signal to new investors.

Selling the holdings to someone else and moving that capital into the new strategy is therefore entirely plausible. But it requires finding buyers. Unlike Kinnevik’s earlier holdings, the names on the list aren’t particularly well known. How about companies like Charm Industrial, Enveda, Nory, Solugen, or Vay? And the ones that have gotten more attention — like Stegra, Aira, and Mathem — have run into serious trouble.

If you’re a Kinnevik portfolio company, you should probably brace for a turbulent stretch ahead. One of your largest shareholders is fed up, and now they’re turning off the money tap.

Looking a year out, several of these oddly-named companies will likely have vanished from Kinnevik’s holdings. Assuming anyone else wants them by then.

This analysis was first published in SvD Näringsliv, in Swedish, on May 25th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

Can Musk Create Another Stock Market Rocket?

May 23, 2026

by Björn Jeffery

Leave a comment

SvD Näringsliv

After the Tesla success, Elon Musk is back. Now he’s taking his rocket company – which today contains more AI than rockets – to the stock market with an astronomical valuation. Can it become another market phenomenon?

When entrepreneur Adam Neumann was preparing to take his co-working company WeWork public in 2019, he had a challenge. How do you make an office hotel sound more exciting than it is?

Neumann decided that the company’s “mission is to elevate the world’s consciousness.” That didn’t quite work. When the world saw the company’s numbers, a backlash began that led to its collapse. And when you read the prospectus for Elon Musk’s rocket company SpaceX, it’s impossible not to think back to WeWork’s fate.

The rocket company’s mission includes, among other things, to “extend the light of consciousness to the stars.” On top of that, a bonus will be paid out if the company manages to build a colony on Mars with at least one million inhabitants.

It sounds a bit like a fever dream. But then again – if there’s anyone who can sell this dream, it’s Elon Musk. He’s done it before, with Tesla’s electric cars.

In many ways, SpaceX has the opposite problem from WeWork. It’s hard to think of anything more exciting than a rocket company. But when you look more closely at the business, you realize that space itself has become an increasingly small part of it.

This is most visible in the section of the prospectus meant to describe the size of the market. In a clearly laid-out bar chart, you find “Space-enabled solutions” – meaning the actual space business. It is the smallest of all categories, valued at $370 billion. Next comes Starlink, which uses satellites to deliver broadband. That market is considered worth more than twice as much.

Those figures are barely visible compared to the “Enterprise applications” category, which appears to cover all software used by all companies everywhere in the world. In this category, SpaceX can offer AI services, through the acquisition of Musk’s company xAI. In total, that market is considered worth $22.7 trillion.

SpaceX is therefore claiming that its total addressable market is roughly the size of the entire US GDP for 2025.

What is called SpaceX is no longer just about rockets and satellites. It is now a company that also encompasses AI development. And that is somewhere around where the IPO finds its logic.

It will take a great deal of money in the years ahead.

The AI business – and the rockets too, for that matter – is losing enormous amounts of money. And it will only get worse. Capital expenditures – fixed assets such as hardware – are running so far this year at roughly seven times higher for AI than for rockets. In the first three months of the year, SpaceX lost more than $4 billion, slightly less than its total loss for all of 2025. The losses are accelerating sharply, apparently to keep pace in the AI race.

All of that might be something you could live with. Elon Musk has done the impossible before. Those who invested early in Tesla have had life-changing returns. Here comes a new company from the same man, and you as a small investor will have the chance to get on board. The list of enthusiastic Tesla shareholders is long – and so is the list of those who regret having missed Tesla’s run.

Needing a lot of money to invest in the future is not, in principle, an obstacle to listing a company. On the contrary, it is often the very purpose of an IPO.

But then we get to the valuation. The exact price won’t be set until just before trading begins, but the signal is clear. The intention is for this to be the world’s largest IPO ever. Talk is that the valuation will be at least $1.75 trillion, possibly as high as $2 trillion.

Sticking to the lower end, SpaceX’s intended valuation sits at roughly 94 times revenue. By comparison, Apple and Microsoft trade at around 10 times, and Nvidia at 21 times. Tesla, which is known for trading far above its automotive peers, is at 15 times.

The valuation is, as you can see, on an entirely different planet. And the reason for that appears to be Musk, AI, and some projected colony on Mars.

If there is any sanity left in the world, SpaceX’s IPO could mark a turning point for this phase of the explosive AI economy. OpenAI and Anthropic are both said to be heading toward their own listings at a rapid pace, while interest in AI remains at record levels. But at some point the market – and perhaps people in general – will say that enough is enough. This is too bubbly, too expensive. We can’t keep going like this.

But any such sanity does not appear to be on the horizon.

Not on this planet – or any other that SpaceX might one day visit.

This analysis was first published in SvD Näringsliv, in Swedish, on May 23rd, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

Insider trading exposes Nvidia

May 20, 2026

by Björn Jeffery

SvD Näringsliv

This analysis was first published in SvD Näringsliv, in Swedish, on May 20th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

Is now a good time to buy Nvidia? Ahead of the quarterly report, that’s the question everyone with a stock portfolio is asking. The answer insider trading gives is unambiguous.

Insider trading has an undeservedly bad reputation. It often sounds like something shady is going on. The label “insider crime” is probably the association that muddies the waters.

But the most common form of insider trading — the kind that isn’t criminal — is where executives at publicly listed companies report their transactions. Purchases and sales. And through these, you can sometimes make out a story about the company in question. Ahead of Nvidia reporting its quarterly results late Wednesday evening, SvD examined precisely that story — and it is both crystal clear and unusual.

Colette Kress, CFO of Nvidia, cannot have been bored at work lately. When she started in 2013, the share price was around $0.40. Today the equivalent figure is around $225 — a record high in the company’s history. Several executives at Nvidia receive shares as part of their compensation — including Kress, who has ultimate responsibility for the company’s finances. After such a run, you can understand the impulse to secure your personal finances by selling some shares.

And her personal finances are well secured today. To put it mildly.

Over the past three years, Kress — an insider at Nvidia — has sold shares worth more than $160 million, over SEK 1.5 billion at today’s exchange rate. As recently as this March, she sold shares worth just under $11 million.

Kress is not alone. An analysis I conducted shows that Nvidia insiders have done nothing but sell shares over the past five years. There is not a single insider purchase registered in that period. Neither executives nor board members have put a single dollar of their own money into buying the company’s shares since September 2020 — and even then, it was only one person who bought. At that point the share price was just under $13, in a transaction worth under half a million dollars, according to filings with the SEC. That weighs rather lightly against the selling that followed.

Nvidia’s founder and CEO Jensen Huang has also sold heavily. In just the past three years he has sold around $1.9 billion worth of shares — just under SEK 18 billion. Huang’s shares, like those of several other insiders, have been sold on a pre-set schedule. The purpose is to try to avoid speculation about why an insider chose to sell at a particular moment. The schedule avoids sending unintended market signals. But the quantity of shares sold is still controlled by the insider in question. And most importantly — you don’t actually have to sell your shares at all.

To add some nuance to Huang’s sales, it is worth looking at how many shares he still holds. He owns just over 3 percent of Nvidia’s total shares, a stake currently worth around $183 billion. Over the past three years he has sold roughly 1 percent of that total.

A personal finance advisor would immediately say that Huang’s disposals are on the low side — despite the large sums involved. That large a proportion of your wealth generally should not sit in a single asset. On the other hand, we are well beyond the argument of securing Huang’s personal future here. We are talking about billions of dollars in sales in just the recent period.

Ahead of Wednesday evening’s quarterly report, the market will be looking for signs of whether the rocket the Nvidia share has become can continue to climb. The pressure from AI companies suggests that demand far exceeds the supply of Nvidia’s valuable chips.

At the same time, it is remarkable that those with the most direct insight into the question — all insiders in the executive team and in the company’s top tier — do not appear to share that view. Why is everyone selling? Why has not a single senior executive bought a single Nvidia share in the past five years, despite this front-row view of the ongoing boom? Even if you receive shares as compensation, you could simply hold them rather than sell.

Insider trading does not give you all the answers about a company’s future. But it raises a number of interesting questions. In Nvidia’s case, the insider sales may stand in contrast to the hype the company continues to generate. Those on the inside are not spending their own money to buy shares. But they are quite happy for others to do so. There is something to take from that.

Share prices in this piece have been adjusted to account for Nvidia’s stock split.

Was the battery purchase a cover?

May 13, 2026

by Björn Jeffery

SvD Näringsliv

This analysis was first published in SvD Näringsliv, in Swedish, on May 13th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

Why did Lyten ultimately buy Northvolt? New information suggests the battery factory was only part of the value. Access to cheap green electricity proved decisive — and it may be routed to an American tech giant.

For a battery factory, there has been a surprising lack of discussion about actual batteries when it comes to Northvolt. Perhaps because relatively few of them ever left the factory while it was running?

A new book raises questions about whether things will improve under new management.

In the aftermath of the Northvolt collapse, it looks as though access to green electricity — rather than battery manufacturing — is one of the primary assets on offer. And moreover, it was what tipped the scales in favor of the American company Lyten’s rescue of Northvolt. Behind this obscure company, you also find considerably more well-known names that were part of the deal.

SvD has previously reported that Lyten — as part of purchasing Northvolt — acquired a new neighbor in Skellefteå: the American data center company Edgeconnex. That company is in turn owned by the Swedish private equity firm EQT.

Edgeconnex plans to build a new data center — potentially the largest in Sweden — on the land right next to Northvolt. Daniel Ketema, communications director at EQT, told SvD in February that “the data center has nothing to do with Lyten, beyond the geographic proximity. These are two completely separate operations.”

The new book “Northvoltfallet” by business journalist Gunnar Lindstedt suggests this may be a truth with some modification.

The book describes how EQT extended Lyten a bridge loan of SEK 2–3 billion to enable the purchase of Northvolt’s battery factory for SEK 900 million, plus real estate and land for roughly the same amount.

Why would EQT have an interest in helping Lyten with the financing? Because, as part of this circular deal, Edgeconnex will gain access to a portion of the green electricity that Skellefteå Kraft had previously guaranteed to Northvolt, as well as the adjacent land. This is according to sources in Lindstedt’s book. SvD approached EQT for comment, but the company has not responded.

According to the book’s sources, Edgeconnex will pay around SEK 5 billion to secure both the land and the electricity agreement from Lyten. That would allow the loan to EQT to be repaid. The circle is closed. The operations are entirely separate — but the original deal could not have happened without the initial loan.

What the deal highlights is the true value that players — old and new alike — are trying to capture in these green tech transactions. It’s about cheap green electricity. But when it comes to electricity access of this scale, you can’t simply plug into a grid and get going. It requires planning and permits.

Being able to use existing electricity agreements already in place is therefore a substantially faster and smoother solution. Skellefteå Kraft was ready for Northvolt. And now some of that allocated electricity may go to a data center instead.

Lyten seems to have developed a taste for the model. In March they announced they were establishing a “Lyten Industrial Hub” in Poland. It is being built — just as in Skellefteå — next to Lyten’s (formerly Northvolt’s) battery factory in Gdansk. Lyten CEO Dan Cook said in a press release that the purpose was to “combine advanced materials and battery energy storage with digital infrastructure for AI.”

But what AI data centers need right now is not large quantities of batteries. They need electricity. A great deal of electricity. And Lyten’s battery manufacturing in Northvolt’s former factories has not yet begun.

As Northvolt is now being rebuilt, it is worth examining what was promised from the start, what was actually delivered — and what the operation looks set to become.

What was promised were Swedish environmentally friendly batteries that would, among other things, supply European automotive manufacturers. A strategically important component to have close at hand in a geopolitically complicated world.

What we got was a Swedish factory with Chinese machinery, operated by Chinese workers, using imported Chinese cathode material. The Swedish cathode that was meant to be produced simply could not be made to sufficient quality or in sufficient quantity. And it all ended in a bankruptcy described as the largest in Sweden since the Kreuger crash.

What does it look set to become? The very thing that was supposed to make these batteries green in the first place — access to electricity in northern Sweden — may now go to an American data center instead.

The end customer — those who intend to use the data center once it is built — is, according to Lindstedt’s book, a familiar name: Google. When politicians celebrate Northvolt’s survival, they would do well to update their picture of what the operation may actually become.

Instead of a major push for green batteries, we may also end up simply helping an American tech giant with its electricity supply.

You can do that. But it is very far from what was promised at the start.

Thousands fired because of “AI”

May 10, 2026

by Björn Jeffery

SvD Näringsliv

AI is frightening corporate executives – and forcing them to make layoffs. That’s what they want us to believe, anyway. So far, it looks more like a convenient excuse than a revolution.

The CEO looks concerned. “We have made the difficult decision to reduce our workforce,” he says. We’ve heard it before.

Variations on this theme are now everywhere in the business world – at tech companies especially.

Coinbase is cutting 14 percent of its staff, Cloudflare is trimming 20 percent, and Meta is laying off a tenth of its employees. But pointing at AI is easier than taking responsibility for a business that isn’t performing well enough.

The public explanation for why these thousands of jobs are disappearing always seems to be the same. It’s AI. Sometimes the company needs to invest more in AI, sometimes AI has disrupted operations, or AI has made it possible to become more efficient.

Either way, it’s AI’s fault we ended up here.

This week, Brian Armstrong, CEO of crypto company Coinbase, wrote a similar explanation for the company’s upcoming layoffs. But in the paragraph before, you could sense something else was going on. “Crypto is also facing the next phase of broader adoption,” he wrote.

A few days later, the company’s quarterly report arrived, and it became clear that the phase Armstrong is in now is something of a minor catastrophe. The company posted a large loss where analysts had expected a profit. Transaction revenues – the very foundation of Coinbase’s business as a trading platform – came in nearly half a billion kronor below expectations. The company – and the crypto market as a whole – has serious problems.

When the underlying business is limping along, it’s easier for executives to blame AI. The numbers don’t support that thesis, however, according to a new report from the Swedish Trygghetsrådet.

Between 2023 and 2025, 50,000 laid-off white-collar workers in the private sector were analyzed. So far, there are no patterns suggesting that industries or companies with high AI exposure have laid off more people than others.

Individual tasks can be affected by AI – jobs change – but the much-discussed wave of AI-driven layoffs can’t be found in the data. Even in the US, where the AI boom is biggest, unemployment is roughly the same as it was six years ago. Before ChatGPT stirred things up.

So what is actually happening? Here are three concepts that might explain what’s going on.

The first is what’s known as the Jevons paradox. It comes from a 19th-century British economist who showed that increased coal efficiency didn’t reduce demand for the resource – it increased it. The more efficient coal use became, the more coal was needed.

The same principle may apply to AI. Rather than replacing workers, AI may end up requiring more of them to manage it – because AI enables a kind of productive work that wasn’t previously possible. Microsoft CEO Satya Nadella has billions of reasons to be biased here, but he has said he believes AI is going through the Jevons paradox right now.

The second concept is Parkinson’s Law. It holds that work expands to fill the time available to complete it. Everyone has experienced this, at home as much as at work. How long does it take to clean the house? If guests are arriving in fifteen minutes, it takes fifteen minutes. Otherwise you can wander around fiddling for hours.

In an AI context, this means that while AI handles many tasks for you, you simply work on something else instead. Perhaps the remaining work now takes longer, or you’ve taken on new tasks that weren’t done before. The upshot: you’re still working full days, even with AI there to help.

The third and final concept is the simplest. Many companies aren’t doing very well right now. The global economy and stock markets are propped up by a handful of enormous tech giants, data center construction, and the promise of an imminent AI revolution. But what doesn’t show up in the indices is that a lot of businesses are struggling. And that, in turn, means layoffs happen from time to time.

Companies have struggled and laid off staff in every era. But blaming AI is easier than admitting you’ve run the business poorly. So sure, AI does serve a purpose in the workplace – it’s our era’s scapegoat. Just not a revolution in the labor market. At least not yet.

This analysis was first published in SvD Näringsliv, in Swedish, on May 10th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

The criticism is right — but Stenbeck is pointing at the wrong target

May 6, 2026

by Björn Jeffery

SvD Näringsliv

This analysis was first published in SvD Näringsliv, in Swedish, on May 6th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

Kinnevik’s strategy was sharply criticized by Cristina Stenbeck at the company’s own annual general meeting this week. But those comments should reasonably have been self-criticism.

Annual general meetings tend to be fairly sleepy affairs. A lawyer reads out a lot of procedural points, most shareholders have cast their votes by post before the meeting starts — and afterwards there’s coffee at the listed company’s expense.

That kind of AGM was evidently not what Kinnevik’s chairman Cristina Stenbeck had in mind. Change was to be signaled.

“The board’s view is that the previous strategy has been too risky,” she told the attending shareholders. “What’s needed instead is more cash flow-generating companies as a base for the business.”

The larger — unanswered — question hung in the room. The share price has fallen more than 78 percent over five years. How did Kinnevik end up here in the first place?

To answer that, we need to look at the basics of how Swedish corporate governance works. Sometimes people oversimplify with “the board governs and the CEO executes.” In practice, the CEO often has considerably more say than that. But the board holds the ultimate trump card: the ability to dismiss them. The question of whether the company has the right CEO — and thus the right strategy — is something the board should continuously ask itself.

But there is another level, and it is particularly important for Kinnevik in the situation they now find themselves in.

The owners — via the AGM — appoint the board. And among the larger and most influential owners, we find chairman Cristina Stenbeck herself.

When Kinnevik’s obviously failed strategy is reviewed, this dimension needs to be included. Kinnevik’s former CEO, Georgi Ganev, did not execute the strategy in a vacuum. It was approved by the board, and continued for several years.

Ganev took over as CEO in the summer of 2017, and the sale of Tele2 — exactly the kind of cash flow-generating holding Kinnevik is now seeking — wasn’t completed until 2024. Where was the oversight then? Where were the owners?

They wouldn’t have had to look very far to find inspiration for the model Stenbeck is now proposing. She talks about “building a backbone” for Kinnevik — that is, more stable companies that don’t bleed money and can finance new investments — which is exactly how Kinnevik used to operate. Companies like Billerud Korsnäs (sold in 2013) and Tele2 served precisely that function in relatively recent memory.

Beyond this new backbone, the investments in climate technology are to be wound down, an internal organizational review is to be conducted, and investments in new companies are to be paused. Furthermore, the company should be selective about investments in existing holdings.

What’s left? We have an investment company that currently has no intention of making any new investments. Even if you’re already in the existing portfolio, it seems you can’t necessarily rely on backing.

The clearest thing to emerge was that some of the company’s own staff would be let go, and an acquisition would be sought to replace the cash flows lost with Tele2. It’s not the most inspiring action plan anyone has ever heard.

The board’s most important task, as noted, is to ensure the company has the right CEO. Now they must prove they are capable of precisely that task. Market confidence is eroded, much of the money has already been distributed to the owners — and the share price just keeps falling. An optimist might say it can only go up from here.

What’s needed is a visionary optimist. Someone who can look beyond Kinnevik’s weak portfolio as it stands, and set a course toward something new. A destination where the share price moves in the right direction and Kinnevik’s position within Swedish business life can be reclaimed. It will take a while.

The situation is tough — but it has had a silver lining for some. One of Sweden’s leading investment companies has been gradually dismantled while its owners collected dividends worth billions.

Receiving generous dividends is one part of what comes with being a major shareholder.

Now, the rest of the shareholders would like to see the other part too — governing responsibly and taking accountability.

An enormous success? Quite the opposite

April 27, 2026

by Björn Jeffery

SvD Näringsliv

This analysis was first published in SvD Näringsliv, in Swedish, on April 27th, 2026. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

The Swedish tech industry sees the multi-billion sale of Cursor as an enormous success. Beyond one individual Swede’s personal finances, it should reasonably be exactly the opposite.

It did not take many minutes before Stockholm was jubilant over the coding tool Cursor’s proposed sale to SpaceX. Yet another victory for the Swedish ecosystem of tech companies? It is rather a masterclass in confirmation bias. People see what they want to see — and not much else.

What has happened is this: an American company is buying another American company, whose one co-founder is Swedish — but he doesn’t even work there anymore. The reason for the sale is, however, interesting to look at more closely — and it is not as rosy as it first appears.

Let us begin by describing Cursor. It is not every day a company that most people have barely heard of is proposed to be sold for 550 billion kronor. The company is actually called Anysphere, and its product — Cursor — helps developers write code using AI. To do this, they use a range of different AI models from companies such as OpenAI, Anthropic, and Google. Relatively recently they also launched their own model, Composer, as a way to reduce their dependence on the giants.

It is precisely that dependence — and its costs — that has put Cursor in a tricky situation.

This week The Information revealed that Cursor has, until now, had negative gross margins. For those not entirely comfortable with financial terminology, this can be explained more simply. When a company takes its revenues and deducts the cost of being able to deliver the service — that gives you the gross margin. In Cursor’s case, they lose money even there. After that, a host of other costs are added — marketing and product development, for example. This means that for every krona Cursor sells for, they lose more than a krona. That model is, for understandable reasons, tricky to sustain for too long.

For companies that are growing rapidly, however, this is not entirely unusual. In theory, the problem should resolve itself as sales increase and the company grows. Many startups have gone under while waiting for this breakeven point. And even if you are on the way to that point, it costs a great deal of money before you get there. This also applies to AI giants such as Anthropic and OpenAI, which are nowhere near profitability.

Cursor was thus out on the market trying to raise several new billion dollars in financing. They received some offers, but also encountered a degree of scepticism — despite the company’s popular product.

Then the fundraising became a sale instead.

The investors’ concern stemmed from Cursor’s supplier, and now competitor, Anthropic and its product Claude. When you use Cursor, it is often Claude doing the heavy lifting in the background. Cursor is an intermediary layer between the end user and the AI services beneath, much like Swedish Lovable, for example.

The prospective investors worried about the possibility that customers could bypass Cursor and just use Claude instead. Which many already do today. If Claude were also to decide not to allow Cursor to use their service, problems could quickly arise.

It was at this point that Elon Musk’s SpaceX stepped in.

They recently merged with Musk’s other company — xAI — which had in turn merged with the social network X (which Musk also owned). SpaceX will buy Cursor for 550 billion kronor, or pay around 92 billion kronor in a fee to Cursor if the deal is not completed for some reason. With the rocket company’s imminent stock market listing — and its reportedly galactic valuation of around 16,000 billion kronor — the price tag is relatively manageable.

For the venture capitalists who invested in Cursor, the deal provides a good return. Selling portfolio companies at high prices is the entire business model. The Swedish co-founder Arvid Lunnemark from Malmö will almost certainly become a billionaire in the process. Like Joel Hellermark, who sold Sana to Workday for 10 billion kronor, the timing is elegant. Interest in AI, and the valuations that follow, has never been higher.

Looking at the value for the Swedish tech ecosystem, it is not obvious that one should celebrate the deal. Cursor was sold because the alternatives to continuing were becoming very limited. The company has never made a single krona in profit, and they burned billions of dollars before reaching this point.

They managed to get out of their precarious financial situation before it was too late. But once again, a large American company is gobbling up a smaller one. One of the few conceivable challengers to the existing oligopoly of AI companies disappears. And one with a Swedish connection at that.

After the Cursor deal, a Sweden-born founder living in San Francisco will become extraordinarily wealthy. Hats off for that. But if anyone thinks this is something to celebrate in Stockholm, they have probably missed the point.

0,75x speed + 3 cultural gems & 3 conversations

April 26, 2026

by Björn Jeffery

Newsletters

In the make-shift podcast studio I have in my office, there’s a small paper note pinned to the wall.

It reads: “0,75x speed”.

It’s a reminder of some advice I got from my friend Magnus. Talk slightly slower, take your time, and don’t get ahead of those who are listening.

“0,75x speed” serves well as a metaphor, far beyond our podcast.

As AI enables us to do more – and faster – the emphasis is leaning heavily on output and efficiency. I get caught up in this myself sometimes too. Feeling productive is somewhat addictive. Listening to sped-up podcasts, getting summaries of books, letting the AI build projects for you while you are sleeping.

But what is “productive” in this sense? If it didn’t need to be done from the beginning, does doing it anyway still count?

Another way of thinking about this newly found efficiency is going in the opposite direction. What can you do, that the AI can’t? How you can shape a point of view that a large language model never could?

I would argue that books are not for summarizing. Even skimming them will give you more than an AI-generated summary. And travelling to places will offer a perspective no LLM will give you, however fast and precise it may become.

As the world seems to speed up, a way to manage it could be doing the opposite. Doing what only you can do – in the way only you can do it.

/Björn

Three cultural gems

Flesh – David Szalay
A book that doesn’t describe anything of what almost every other book in the fiction genre covers. No inner monologue, no thoughts. Just flesh. A fascinating read. And a Booker prize winner, which should be enough to make you curious.

Oops – Momo Boyd
This new song flirts with 2000s R&B but still feels contemporary. The whole setting – new album + Colors video – feels the making of a new star.

Trainjazz.com
This is what the internet was made for. Real-time data from subway trains in New York is translated to a piece of jazz that changes depending on the movement. And if you’re close to certain trains, their notes grow louder. A work of art.

Three conversations about the books of Silicon Valley

Together with (my friend and) historian David Larsson Heidenblad, I did a series of conversations centered around the ideas, knowledge and history of Silicon Valley. Starting with a variety of influential books, we went through what can be learned from them, how these ideas have been used in practice, and how you can use them yourself – without having to run a startup.

Click the links above to listen to the conversations, if you’re curious – and a full book list is here.

Originally published on Substack on April 26th, 2026.