In this week’s edition of The Prototype, we look at whether AI can withstand economics, using quantum computers for bond trading, a transistor revolution, and more. To get The Prototype in your inbox, sign up here.
Interior of a Facebook data center
AFP via Getty Images
Tech companies are determined to come out on top in the AI race, no matter the cost. And that may pose a long-term problem, according to a new report from Bain & Company, which finds that the AI’s computational demands are growing at about twice the speed of Moore’s Law. Which poses another particular challenge because Moore’s Law itself, which predicts that computing power per dollar doubles about every two years, is slowing down as chip manufacturers run into physical limitations.
No wonder, then, that the number of data centers is growing fast in order to keep up. But that growth faces a serious barrier: economics. Bain’s report finds that by 2030, tech companies will need to find $2 trillion in new AI revenue to make a profit on the cost of deployment–but estimates that they’ll be about $800 billion short.
What does that mean for the economy as a whole? That’s an important question. A research note from Deutsche Bank from earlier this week suggests that without spending on the infrastructure for AI, the U.S. economy would likely be in a recession right now. And if companies can’t find the revenue they need to keep spending that money, it could have ripple effects well beyond Silicon Valley.
Stay tuned.
HSBC and IBM Used Quantum Computers To Predict The Bond Market
A team at HSBC bank, working with researchers at IBM, was able to successfully solve real-world bond trading problems using quantum computers.
Today, institutions around the world use algorithms to make trades, which are pre-programmed to account for variables like price and volume to best meet customer orders. Data scientists are constantly working to tweak and improve these algorithms to optimize trading strategies for their clients. A crucial aspect of that is predicting future prices in order to maximize profitability.
The team at HSBC wanted to know if quantum computers–which are ideal for optimization problems–could help. Working with IBM, the researchers developed a hybrid approach that combined conventional and quantum algorithms, the latter of which were run on IBM’s quantum computers. The result was a 34% improvement in prediction accuracy compared to purely classical computing approaches.
Although more work needs to be done to refine the approach and utilize it in a generalizable way, this represents a “tangible example of how today’s quantum computers could solve a real-world business problem at scale and offer a competitive edge,” Phillip Intallura, HSBC’s head of quantum technologies, said in a statement.
DISCOVERY OF THE WEEK: MAGNETIC TRANSISTORS
Engineers at MIT have created magnetic transistors, which could enable faster and more energy-efficient semiconductors. Researchers have been trying to use magnets this way for years, but the materials used so far haven’t been optimal for computing functions. That changed after experimenting with chromium sulfur bromide, which replaces the silicon in a conventional microchip and enables the transistors to be switched on and off with an electric current. More interestingly, magnetic transistors allow for memory to be built into the chip instead of as a separate component, which could speed up a number of applications, especially AI. For their next step, the researchers will focus on scaling production.
WHAT ELSE I WROTE THIS WEEK
In my other newsletter, InnovationRx, Amy Feldman and I looked at the medical community’s response to Trump’s Tylenol claims, a startup working to fix drug pricing, Pfizer’s obesity bet, using AI to manage clinical trials and more.
On Forbes’ Breaking News YouTube channel, I sat down with my colleague Maggie McGrath to go more in depth about Tylenol, autism, vaccines and other medical claims the Trump Administration made this week.
SCIENCE AND TECH TIDBITS
Impulse Space signed a 2027 launch deal with satellite communications company Astranis. Under the agreement, a SpaceX Falcon 9 will deliver Impulse’s Helios rocket to low Earth orbit. From there, Helios will launch into geostationary orbit, where it will then deploy Astranis’ satellites in less than 24 hours.
Venture capital firm Acumen has raised nearly $250 million for its effort to electrify 70 million people in sub-Saharan Africa, using primarily off-grid solar power.
Major League Baseball is bringing robot umpires to the plate, which will determine a ball or strike in the event of a challenge to the human umpire’s call.
NASA announced its newest class of astronauts: 10 people selected from over 8,000 candidates. The agency’s latest explorers will join the existing astronaut corps, and for the first time ever, a majority of the class are women.
PRO SCIENCE TIP: GIVE EMPLOYEES FRIDAYS OFF
If your employees are stressed and burning out, consider restructuring to a four-day work week. That’s the conclusion of a recent, large-scale study which focused on nearly 3,000 employees across 141 organizations in six different countries. Before embarking on the experiment, the companies spent eight weeks restructuring workflows–and trimming inefficiencies like unnecessary meetings. The researchers then followed the employees for six months of four-day weeks. The result? Employees reported better mental health, less fatigue and higher job satisfaction. And while the study didn’t measure productivity, here’s one sign a four-day week makes things better for companies: After concluding the study, more than 90% of participating companies opted to keep it.
WHAT’S ENTERTAINING ME THIS WEEK
I recently caught The Long Walk in theaters. Based on the Stephen King book of the same name, it’s set in a dystopian United States where once a year, young men are drafted into the titular contest: forced to walk non-stop until only one remains. The strength of the film comes from the bonds that form between the contestants as they’re slowly dispatched one by one. There are some great young actors here, including Cooper Hoffman and David Jonsson both giving star-making performances. This is easily a top ten Stephen King adaptation.