Bitcoin ETFs
Coin of the realm
The SEC was right to let spot bitcoin ETFs trade, but it should stop pretending a court did not force its hand.
The money behind technology: capital, markets, margins and the firms that move them.
Bitcoin ETFs
The SEC was right to let spot bitcoin ETFs trade, but it should stop pretending a court did not force its hand.
AI and young workers
A Stanford study finds that young workers in AI-exposed jobs are being hired less, which is worth taking seriously, but the evidence is too thin to justify rules against the technology.
Initial public offerings
Figma's dazzling debut shows that public investors want good startups, and that bankers still struggle to price them.
Thiel Fellowship
Peter Thiel has doubled the price of dropping out just as AI makes small teams mighty, yet students are drifting back to campus, and on the evidence they are right to.
Reddit's IPO
Reddit was right to let its users buy in, but a flotation that leaves control with insiders is a poll in which only some votes count.
Instacart's IPO
Instacart's listing proves that the window for technology flotations is open again, but only for firms willing to accept what the market will pay.
Silicon Valley Bank
Washington was right to protect SVB's depositors, but startups should learn the lesson that rescue cannot teach them.
Skipping university
Americans are leaving college and starting firms at record rates, and cheap AI will make that trade more tempting, but a degree remains the better bet for most would-be founders.
ChatGPT in schools
New York's ban on ChatGPT protects nothing it claims to protect, and schools would do better to teach pupils to catch the bot's mistakes than to pretend it is not there.
FTX
FTX's collapse needs no new crypto law, only the enforcement of old rules about whose money is whose, and a little more suspicion from investors.
Venture capital
Startups have raised more in nine months than in any full year before, and the boom rests on a price of money that is about to change.
Coinbase's listing
Coinbase has shown that a company can go public without paying a bank to guess its price, but the first day showed how much guessing the market itself does.
SPACs and the SEC
The SEC staff's warrant ruling is a footnote with teeth, and the SPAC boom should be allowed to cool under it.
Retail trading
Robinhood's curbs on GameStop were forced by plumbing, not conspiracy, but a market whose settlement lags its apps has no business promising instant access.
Quibi
Quibi's failure is not a verdict on short video but on a business that spent $1.75bn before asking anyone what they wanted.
Campuses close
The class of 2020 is learning that a degree bundles a lecture with a place and a crowd, and universities should now say which parts of the bill they are still delivering.
WeWork
The collapse of WeWork's flotation is not a failure of public markets but their finest hour: investors read the prospectus and declined.
Student founders
Student debt makes skipping university look shrewd, but the best evidence says successful founders are closer to 45 than 19, so the case for dropping out is narrower than its fans claim.
CoreWeave and the AI-cloud trade
CoreWeave's cut-price flotation is a fair test of whether renting out Nvidia chips is a business or merely a boom, and one customer is sitting the exam for it.
DeepSeek and Nvidia
A cheap Chinese model has rattled the AI trade, but it makes computing cheaper rather than pointless, and the bill for the boom will still be paid
Intel's chief executive
Pat Gelsinger's departure leaves Intel with the same problem and two temporary people to solve it; the board should use the interval to split the firm.
OpenAI's funding round
OpenAI's investors are paying a record price for a company that does not yet make money, and have written a fallback into the contract.
Arm's flotation
Arm's return to the stockmarket is a fair price for a great business, but SoftBank is selling a toll road with a lot of its traffic in one lane.
Nvidia
Nvidia's guidance proves the AI boom is real, but no single supplier should be priced as if it owns the whole of it.
Nvidia and Arm
Regulators were right to stop Nvidia buying Arm, but the next owner of a chip standard should be one that every rival can trust.
AMD and Xilinx
AMD is right to pay in its own richly valued shares for Xilinx, and regulators should let a deal that makes Intel's life harder go through.
Alphabet's capital return
Alphabet's first dividend says less about a maturing firm than about a cash pile that outgrew its owner's ideas for it.
Microsoft's market value
Investors have priced Microsoft as the winner of the AI race, yet the prize still depends on customers paying $30 a month for a chatbot.
Meta's year of efficiency
Investors have rewarded Meta for spending less, which is sensible, but efficiency is a means and the firm has yet to say what it is for.
Twitter and Elon Musk
Mr Musk has paid full price for a loss-making firm whose customers can leave at will, and no change of ownership will fix that.
Broadcom and VMware
Broadcom's $61bn price for VMware is hard to earn back without raising bills, so regulators should ask whether customers can walk away.
Amazon's succession
Jeff Bezos has picked the man who built the firm's profits, and the choice says more about where Amazon earns its money than about who will run it.
Salesforce and Slack
Regulators should wave through Salesforce's $27.7bn purchase of Slack, which adds a serious rival to Microsoft; the price is a matter for Salesforce's shareholders.
Microsoft and OpenAI
Microsoft's $1bn bet on OpenAI is less a punt on thinking machines than a prepaid order for its own cloud, and that makes it a shrewd one