A daily market update from FS Insight — what you need to know ahead of opening bell

“You can’t stop progress, but you can help decide what is progress and what isn’t.” — Ashleigh Brilliant

Overnight

Argentina’s record $50.3 billion bond swap aims to pave way to ending currency controls 

Wall Street slams NY lawmakers over sovereign debt bill

Oil prices jumped over 2% after Ukraine’s strikes on Russian refineries

Stripe’s payment volume surged 25% to $1 trillion last year; business was cash-flow positive in 2023, revenue-automation now a ‘giant part’ of it

Gerald M. Levin, cable TV visionary and Time Warner chief, dies at 84

German military still faxes documents, keeps medical records on paper, and cannot radio allies, official warns

Bill that would send billions to high-speed rail gains support in the House

IRS’s new Direct File system is under experimentation in 12 states; 19 million taxpayers could be eligible to file their taxes with the IRS at no cost

A new study found that efforts to reforest the eastern United States have cooled it down

Hedge fund leverage is near record levels

Yellen says U.S. rates ‘unlikely’ to return to pre-Covid levels

SPAC Porticoes Capital plans to acquire failed US banks

Goldman to resume bets on U.S. property, despite peers pulling back

Brookfield says U.S. office market is world’s most oversupplied

Goldman private credit veteran and co-head Lee Becker makes an exit

EY CEO-elect Janet Truncale to shuffle leadership and plan cost cuts

PwC Australia let go of 37 partners, ~300 others following tax leaks scandal

PIMCO filed for its first mutual fund-to-ETF conversion

Toyota agrees to biggest wage hike in 25 years.

First news

• The AI industry tries to compensate for its own hype
• The U.K. launches an AI-at-one-thousandth-of-the-cost initiative
• A new startup called Cognition allows users to build code from scratch
• Microsoft launches a build-a-chatbot function and a Copilot security product
• China introduces AI for surgeons
• The founder of Google DeepMind points to AI’s ability to extrapolate how to play all games by knowing how to play one
• The E.U. lays down major AI regulation.

Charts of the Day

MARKET LEVELS

MORNING INSIGHT

Good morning!

Did auto leasing rates explode higher at a 40% annualized rate in Jan and Feb 2024?

Nope… yet, this is what the CPI is showing. 

A regional bank rally suggests the markets are sniffing out a more dovish Fed.

More in our Macro Minute video, linked here.

TECHNICAL

We continue to see the U.S. stock market as being attractive, technically speaking, and do not feel sufficient risk is there to warrant a selloff at this time. While momentum gauges like RSI have gotten overbought, there remains precious little other evidence with regards to frothy speculation to the tune of excessive valuation measures that would warrant a major selloff. Powell’s recent dovishness likely won’t change too dramatically despite a second “hot” CPI reading, and it appears like bounces in U.S. Dollar and Treasury yields could prove short-lived before weakness back to new monthly lows into this summer. Rallies up to SPX-5250-5300 look possible ahead of a possible late-March pullback into April. However, without any evidence of this, it pays to wait for signs of weakening breadth and/or momentum before expecting a selloff of any sort.

Click HERE for more.

CRYPTO

• Spot Bitcoin ETFs recorded 14,706 BTC, or over $1 billion, in net inflows on Tuesday, eclipsing the previous February record of $673 million. BlackRock achieved a record $849 million in inflows, whereas Grayscale experienced a relatively modest $79 million in outflows, significantly below its recent trend. This caps a recent stretch from Monday, March 4th, through Tuesday, March 12th, during which net inflows into spot ETFs totaled over $3.8 billion.
• Additionally, capital flows on the stablecoin front have been decidedly positive. The total market cap of stablecoins is approaching $146 billion, indicating a rapid influx of capital into the crypto ecosystem.
• Coinbase announced plans to issue $1 billion in Convertible Senior Notes due 2030 in a private sale aimed at qualified institutional buyers, with an additional $150 million option for over-allotments. These notes, unsecured and accruing semi-annual interest, may be converted into cash, Coinbase Class A stock, or a mix, at Coinbase’s discretion. The specifics, including the interest rate and conversion terms, will be determined at the time of the offering. To mitigate potential dilution from the conversion of notes and manage cash payment obligations, Coinbase plans to engage in capped call transactions with certain initial purchasers or financial entities. These agreements are designed to limit stock dilution and cash payments above the principal amount upon note conversion. Coinbase expects that these transactions might influence its Class A stock price due to initial hedge setups and subsequent market transactions by the counterparties. The company plans to use the sale proceeds to manage and repay existing debts, support general corporate activities, and possibly invest in or acquire new companies, products, or technologies.

Click HERE for more.

FIRST NEWS

Pumping the Brakes. In the past year, major technology firms have talked up generative artificial intelligence as the next big thing, boosting the stock market to new highs. Yet behind the scenes, representatives of major cloud providers and other firms that sell the technology are tamping down expectations with their salespeople in the belief that the hype about the technology has gotten ahead of what it can actually do for customers at a reasonable price. Several executives, product managers, and salespeople at major cloud providers – Microsoft, Amazon Web Services, and Google – also privately confirmed that most of their customers are being cautious. The Information

Music to Our Ears. In related news, the U.K.’s Advanced Research and Invention Agency (ARIA) is launching its first funded program, a $54 million push to develop AI at 1/1000th of the current cost by bringing down compute-hardware expenses. Disbursed over four years, the cash will go to researchers and engineers – also chemists and biologists – to explore how nature efficiently processes information and how that can be applied to AI development.

ARIA notes that humanity has benefited from decades of increasing computing power at increasingly lower cost – but the buck stops at AI. “Our current mechanisms for training AI systems utilize a narrow set of algorithms and hardware building blocks, which require significant capital to develop and manufacture,” it said. “The combination of this significance and scarcity has far-reaching economic, geopolitical and societal implications.” Semafor

Book nook. If your name was Frank McCourt, you could do worse than author a couple of books. Heeding the call with the help of journalist Michael Casey, Frank McCourt-the-living-billionaire is halfway toward accomplishing that goal. In Our Biggest Fight, a new book on what social media has done to our society, McCourt argues that we’ve become the serfs of big-tech companies Meta, Google, Apple and others, which have spawned a digital feudalist system. McCourt’s solution: Decentralized Social Networking Protocol, which helps users take control over their digital identities and data. McCourt says such a move would be good for our well-being and our democracy. Read it yourself – no AI summaries, now – and see if you agree. Semafor

Re-Cognition. Cognition, a new AI startup that came out of stealth Tuesday, is the talk of the automated coding world. AI code generation, a space led by Microsoft-owned GitHub, which offers the coding assistant Copilot to help programmers complete grunt work, was the first commercial use case of generative AI.

Cognition takes the concept a step further with an AI coding tool called Devin, which can build an entire coding project from scratch, fixing bugs and testing code as it does so, making anyone a software developer – something most experts in the field were saying was many years away.

In the meantime, Microsoft has announced that its Copilot Chatbot builder is now available to all Copilot Pro subscribers. Microsoft clarified that it developed the tool without input from OpenAI, perhaps pointing to the effect of recent lawsuits and increasing scrutiny from regulators in the U.S., U.K. and Europe in lessening its reliance on Sam Altman’s company.

If you are a Pro subscriber of the service from Microsoft (priced at $30 per user per month), you can click a new menu labeled ‘See all Copilot GPTs’ that will allow access to a number of GPT’s created by Microsoft alongside the option to create your own. One of the most significant features (one must justify the premium on OpenAI’s $20-a-month charge, after all) allows users to build a chatbot that can pull in data from the organization itself to help with highly specialized tasks, i.e. HR or company guidelines. The builder requires very little technical expertise and uses natural-language prompts for much of the creative process.

The company also announced that it will make its new OpenAI-powered cybersecurity software, dubbed Security Copilot, available on April 1* under a pay-as-you-go licensing model, in a standalone portal. The software will also be sold as an add-on to Microsoft’s existing security products such as its antivirus software Defender and its login software Entra ID. Copilot for Security can process prompts and respond in eight languages. The program also has a multilingual interface for 25 different languages. Semafor, TechRadar, The Information, WSJ

^{* We have no reason to believe that this is a joke.}

Dr. AI. As China backs AI development across the nation this year, Beijing begins testing an AI chatbot for brain surgeons in hospitals. One of many initiatives the government is backing to try and harness the technology, the AI assistant for neurosurgeons is being rolled out at seven hospitals in Beijing and other cities in the coming months. A Hong Kong-based agency under the Chinese Academy of Sciences, the country’s leading state-backed scientific institute, recently introduced an AI model based on Meta Platforms Inc.’s Llama 2.0. Researchers trained and fine-tuned the model with papers, medical journals, and manuals to act as a surgery consultant for doctors. Bloomberg

If You’ve Seen One Game, You’ve Seen Them All. Shane Legg, co-founder and chief AGI scientist at Google DeepMind – whose victory lap so far has included defeating the world champion of the board game Go, revolutionizing biotech with its protein-fold breakthrough, using AI to get us closer to nuclear fusion, and upending the science of weather prediction – has pointed out that as AI models get bigger, researchers are noting the development of crossover skills. The act of learning one language, for instance, allows an AI model to pick up entirely different languages more quickly. You might call it a kind of artificial wisdom. There may be power yet in AI’s narrow ability to generally understand them (winning could come later). AGI might play out via super-intelligence models that, when tasked with a narrow problem, will create software tools to solve it. Kind of like what humans do today. Semafor

E.U. to AI: Beware. The European Union’s parliament approved the world’s first major set of regulatory ground rules for AI yesterday, seeking to govern the technology at the forefront of tech investment. Conceived in 2021, the E.U. AI Act divides the technology into categories of risk, ranging from unacceptable, in which case it is banned, to high, medium, and low hazard. The regulation is expected to enter into force in May, after passing final checks and receiving endorsement from the European Council.

Meeting in San Francisco last year, US and Chinese presidents Joe Biden and Xi Jinping failed to make significant progress on AI co-operation. At the same time, weaponization of AI is already here, as the technology is actively developed and deployed by nation states and private companies; for example, facial recognition and autonomous drones, once theoretical, have been used in Ukraine. Without safeguards, we could see lethal autonomous weapons spread worldwide.

To some extent, there is an ‘arms-race’ narrative in the field of AI, and the likelihood that the worst excesses of science will be mitigated and managed must be measured and encouraged. The UN Outer Space Treaty of 1967 was signed at the height of geopolitical tensions, yet it is predicated on the duty to consult and consider the views of other nations, declaring space “the province of all mankind”. The treaty explicitly links science with the global project of humankind, stating that innovation in space must be for the benefit of all, and creates the conditions for peace in space without hamstringing innovation. As a result, governments have been able to pursue co-operative efforts in telecommunications, in climate science, in shared sensors, and in the International Space Station.

As space was in the mid 20th-century, AI is a new frontier. As in the early days of the space race, AI has become a proxy for opportunity-happy politicians, tech-struck scientists, and shrewd private companies, all eager to use it to further their own narrow goals.

Most AI development currently happens within private companies, while the UN is less powerful than it was. But this should not prevent us imagining, and demanding, a future for the technology that is at least as inspiring and uplifting as the achievements of those who took us to the Moon.

Restrictions, even outright bans, on certain types of weapons have always been necessary. In the case of AI-based weaponry, world leaders should not wait for a disaster. Some may see the Biden-Xi meeting as a missed opportunity – perhaps no one more so than the E.U. CNBC, FT