Dear Aventine Readers, 

There are currently five antitrust complaints filed by the U.S. government (sometimes with states in tow) against some of the world’s largest technology companies. At the heart of most of them, including the latest — filed in March against Apple — is the belief that the companies are depriving consumers of greater choice and better products by stifling innovation through market dominance. But when it comes to technology, fostering innovation through antitrust isn’t always straightforward. This month we look at why the downstream effects of antitrust actions are often difficult to predict and not necessarily what was desired or expected.

Also in this issue: a possible breakthrough treatment for liver disease that prompts a patient’s body to grow healthy new liver tissue; the country’s first electric tugboat; and five experts weigh in on the future of green steel. 

Thanks for reading! 

Danielle Mattoon
Executive Director, Aventine

Subscribe

Subscribe to our newsletter and be kept up to date on upcoming Aventine projects

Your smartphone isn’t as good as it could be, and it’s Apple’s fault.

That’s the fundamental argument from the 88-page complaint made by the Department of Justice against Apple in March. The government, along with fifteen states and the District of Columbia, alleges the company has sought to increase its market power and stifle innovation by “delaying, degrading, or outright blocking technologies.” As a result, it claims, all smartphone users — including iPhone owners — have lower quality devices.

Apple sees things differently. “We innovate every day to make technology people love,” its statement proclaimed. “This lawsuit threatens who we are and the principles that set Apple products apart in fiercely competitive markets.”

This disagreement speaks to the complexity around market power, innovation and the impact that antitrust action has on each. “Underneath it is the basic dispute about how we get innovation,” said Harry First, a law professor at New York University who specializes in antitrust. But this is far from a solved challenge. On the one hand, large, powerful companies command huge R&D budgets that can bring about great technological advances; on the other, competition is often a key factor in product improvement. And legal action to redress imbalances in the market can have effects both good and bad, intended and unintended, on innovation. 

Aventine spoke with experts in antitrust law and innovation to better understand how antitrust cases aim to shape innovation and what this and other suits could mean for future technological progress. They suggested that not all innovation is equivalent, explained why innovation in and of itself isn’t necessarily a consumer good, and speculated about the impact — and potential unintended consequences — of antitrust remedies on the development of new technologies. 

Not all innovation is equal

The complaint brought against Apple by the DOJ is just the latest in a series of antitrust complaints filed against Big Tech companies over the past four years. (There’s one each for Amazon and Meta and two for Google.) 

Central to two of these cases is the way the companies are said to have stifled innovation: Google is accused of preventing rivals from offering new types of internet search; Meta is accused of suppressing the growth of competing social networks. At the heart of such cases is a deep-seated assumption in antitrust theory that competition and innovation are highly related. “Broadly speaking, you would say that a competitive market will probably see more innovation than a monopoly,” said Sruthi Thatchenkery, an assistant professor at Vanderbilt University who specializes in how firms can best compete and innovate in high-tech industries. But it’s possible to argue that this doesn’t quite hold for the technology industry, despite the concentration of power within a small number of companies. “You would never say that the characteristic feature of tech is low rates of innovation. Just the opposite,” said Herbert Hovenkamp, an antitrust law professor at the University of Pennsylvania Carey Law School. He points out that of the companies granted the highest number of U.S. patents in 2023, almost all are technology companies, and Apple, Google and Amazon are all on the list.

So antitrust experts consider different kinds of innovation: sustaining innovation, which incrementally improves existing products, and disruptive innovation, which creates new products and markets. The big tech companies “primarily have sustaining innovation,” said Maurice Stucke, a professor of antitrust law at the University of Tennessee, Knoxville, and co-author of the book How Big-Tech Barons Smash Innovation — and How to Strike Back. “They're not going to purposefully disrupt their own ecosystem,” he added. “But then outside of their ecosystem, they will engage in disruptive innovation.”

The Apple case and Microsoft’s shadow 

It’s the areas where companies are involved in sustaining innovation — often using it to further entrench their dominance — that are targeted by antitrust cases: search and advertising for Google; social networks for Meta and the smartphone ecosystem for Apple. By taking legal action, the theory goes that long-stifled markets can be opened up so that disruptive innovation can flourish. 

In the case of the U.S vs. Apple, the DOJ takes aim at a handful of practices, all of which in one way or another make it difficult for non-Apple players and users to operate seamlessly within the Apple ecosystem. 

Apple will likely argue that any measures it took that may have resulted in a degraded experience for users, app developers or competitors were necessary “for either security or for quality of operation,” said Hovenkamp. It’s far too early to tell how the case will play out in court, because the complaint offers only a glimpse into the reasoning of the Justice Department; we still need to see what evidence it will bring to support those arguments. But, Hovenkamp points out, “every time Apple has asserted those defenses in the past, it has won.”

The complaint against Apple, along with other cases against Big Tech, have been compared to the landmark tech antitrust case of the 1990s that saw the U.S. government sue Microsoft. The software maker was accused of abusing its market power by bundling its browser, Internet Explorer, with its Windows operating system to the detriment of its rivals. Microsoft was found guilty, though it succeeded in not being split into two companies through appeal; it ultimately settled with the DOJ. In the end, Microsoft has had to comply with a number of what antitrust experts call remedies in order to prevent the company from dominating complementary markets; these include injunctions that forced it to enable PC manufacturers to adopt non-Microsoft software.

The question is, did the remedies work? 

“There's an ongoing debate about the success of the Microsoft injunctions,” said Hovenkamp. One problem is that there is often no control group to enable a meaningful comparison of whether or not a remedy succeeded — there’s simply a before and after.

And yet Thatchenkery, along with Stanford University professor Riitta Katila, did manage to study the impact of the Microsoft injunctions on innovation in some markets where the company operated. The pair looked specifically at the universe of products affected by the Windows Server operating system product line — part of the business in which it had dominant products for some parts of the enterprise market but not others. This allowed them to compare areas where the government meaningfully weakened Microsoft’s position in the marketplace against areas in which it continued to operate as before. The research showed that patent activity increased dramatically in parts of the market where Microsoft was now restrained compared to the control group, but also that there was no increase in the number of product launches by its competitors in the newly liberated markets. So innovation in some pure sense may have increased, but not in a way that benefited users.

In hindsight, the Microsoft suit did little to affect the company’s dominance in operating systems: Windows is still the company’s fourth largest revenue stream and holds about 70 percent of the computer operating system market share. Yet it is often argued that the remedies were enough to sufficiently open up the playing field for other forms of innovation. Most of the experts Aventine spoke with shared that view, agreeing that the remedies increased competition in the browser market and provided conditions for new businesses to prosper, including Apple’s iPod (by enabling iTunes to exist on Windows) and, perhaps biggest of all, Google (making it possible for Chrome to exist and gain traction). 

“[But] how much of that was due to the injunction, which opened up the market, and how much of it was due to the rapid expansion of Google?” asked Hovenkamp. “I think we'll probably end up debating that for the rest of our lives.” Either way, said Stucke of the inability to break up Microsoft, “it's interesting that even [perceived] failures in antitrust efforts can create windows of opportunity.”

What might antitrust remedies look like? 

All of the experts that Aventine spoke to agreed that it’s too early to tell what sorts of remedies might be proposed if the suits against Apple, Google and Meta are successful. They did, however, acknowledge that orders to split up the tech companies — an onerous process — are less likely than the imposition of rules upon their conduct, though the Federal Trade Commission did request in its complaint that Meta divest Instagram and WhatsApp.

They also all agreed that devising remedies will be difficult. “Any remedy, if it's cumbersome and isn’t carefully designed, can do more harm than good,” said Stucke. “It could have multiple unintended consequences; it could help preserve the monopoly; it could hamper innovation.” One of the issues here is how focused a remedy is: Thatchenkery explained that if a remedy is overly specific, a company can “just weasel their way round” the rule; overly broad, and it might mean that a company gives up working in a specific area, even stopping work that might be considered beneficial to the market.

Even when a remedy is well designed, it’s almost impossible to imagine all the downstream effects. “I'm tempted to say everything has unintended consequences,” said First. If a company can no longer exploit its market power, it might, say, change policies in other areas of its business to make it harder for third parties to compete. Thatchenkery gave the example of Apple being forced to open up its code so that third parties could make use of the iPhone’s touch-to-pay functions, but the company could simultaneously downgrade its development tools for third-party app developers to make it harder for them to create compelling products.

It’s also possible that remedies could seek change that inadvertently reduces customer choice — a kind of backward step in innovation. One way of thinking about the current smartphone duopoly is that iPhones are a premium product with high security and Android devices are more affordable and have open operating systems. One potential remedy to the charges leveled at Apple might be to force the company to make iOS far more open. “If the only thing this lawsuit accomplishes is to make Apple look more like Android, I would not regard that as a success,” said Hovenkamp.

Then there’s the nagging concern that antitrust action might just be too slow to keep pace with modern innovation. “It's the nature of the beast,” said Stucke. “You have to go through all of these extraneous formalities, all of these requirements, and then to bring the case you're looking at years of litigation, scores of experts, super expense, and by the time it's over any relief might be too little too late.”

The next wave of tech antitrust

It’s against this backdrop that the next wave of anti-competitive behavior in tech may be emerging — advancing at a speed that is taking even some of its developers by surprise. 

Artificial intelligence is predominantly being developed, Thatchenkery points out, by large technology companies — including Google, Meta and Microsoft in collaboration with OpenAI — partly because of the enormous resources AI requires. In a recent paper co-authored with University of Oxford law professor Ariel Ezrachi, Stucke wrote that the current AI ecosystem is at risk of becoming a market where “a few powerful firms can distort the innovation paths, chill innovations that disrupt their ecosystems, and promote innovations that complement their technologies and reinforce their power.” And “failure to move swiftly … may leave us with a repeat of past policy shortcomings.”

Stucke likens litigation against tech companies to duck hunting. “When you hunt for a duck, you're not going to shoot where the duck is, you're going to have to always shoot where the duck is going,” said Strucke. With most attempts to tackle technology through antitrust — including the U.S. vs. Apple case — it’s “sort of like you're shooting where the monopolist was, not where the monopolist is going.”

A spraying machine designed for cloud brightening on the flight deck of the Hornet, in Alameda, Calif. Ian C. Bates/The New York Times/Redux

The first U.S. cloud brightening experiment tests a potential climate fix. At the start of April, on the deck of a decommissioned aircraft carrier docked at Alameda, California, scientists from the University of Washington made history. Using a device known as CARI — or Cloud-Aerosol Research Instrument to give its full title — they sprayed a fine mist of sea salt aerosols into the air in order to observe how the particles would move and change once released. The goal: to increase our understanding of the potential impact of doing such a thing on a much larger scale to create massive clouds that could, in theory, reflect sunlight away from the planet. The New York Times reports that the team planned and performed the experiments in relative secrecy, and for good reason; the practice is highly controversial. The Washington scientists contend that they don’t want to use the technology at scale; rather, they want to understand what its effects might be if humankind runs out of options to cool the planet. The issue is obviously fraught, but it’s undeniable that data would allow us to make a more informed decision about using the technology in the future. (The first issue of this newsletter grappled with the competing viewpoints on a related topic that involves releasing sulfur compounds into the Earth’s stratosphere. One such high-profile project conducted at Harvard was recently shut down; MIT Technology Review has a fascinating piece on the lessons learned from the aborted effort.)

Cell transplants could grow a new liver inside your body. For people with advanced liver disease, there’s often only one option: a transplant. But there are almost 10,000 people on the U.S. waiting list and alternatives, such as transplants from animals, are complex because gene editing is required to make the procedure safe. Wired reports that a biotech company called ​​LyGenesis has a different idea: an injection of liver cells into one of the body’s hundreds of lymph nodes that can help grow new, functioning liver tissue inside the human body. Early tests in mice and pigs with liver failure have shown that the technique can be used to grow liver tissue complete with blood vessels and bile ducts, adding to the animals’ overall liver function and compensating for the damaged organ. The company recently kicked off trials in humans, injecting 50 million liver cells harvested from a donor into a single lymph node in a patient with end-stage liver disease. By choosing a lymph node close to the liver, the researchers hope that the new mini livers will benefit from molecules emitted by the original organ to stimulate regrowth. If early experiments are successful, future patients will receive injections into three to five lymph nodes. The company says the treatment costs a fraction of what a traditional liver transplant does, so if effective it will be transformative for people with liver disease and their families. A final note: Anyone receiving this treatment will need to take immunosuppressant drugs indefinitely due to the presence of donor cells in the body. 

Factories to build the chips of the future are coming. In mid-April, the Commerce Department announced that Samsung will receive $6.4 billion in funding, an addition to Samsung’s own investment of $40 billion for its Texas semiconductor facilities, which will include a so-called advanced chip packaging facility. Earlier that month, memory chip maker SK Hynix announced that it plans to build its own $3.87 billion advanced chip packaging facility in West Lafayette, Indiana. Increasingly, the semiconductor industry is turning to the process of something called advanced chip packaging, which is essentially merging different computer components, particularly processors and memory, to eke out performance gains in semiconductor chips. (The Financial Times has a neat visual explainer.) By positioning chips and memory together on the same piece of silicon — or even, in something called 3D packaging, by stacking several layers of memory directly on top of a processor itself — there can be more, and shorter, paths for data to travel, allowing the components to operate at greater speed and efficiency. Up until now, such gains were achieved by the ability to manufacture ever-smaller transistors, the minuscule building blocks for chips; the more transistors on a chip, the more efficient and powerful that chip is. But modern transistors are now made at sizes close to the atomic scale, which means we’re reaching the physical limits of how small they can be. Thus the need for alternate ways to improve chip performance. The recently announced facilities pave the way for a new generation of chips to be manufactured on U.S. soil, and help the nation nudge closer to the Biden administration’s aim of having the country produce 20 percent of the world’s semiconductors by 2030.

Steel production in an electric arc furnace. Norenko Andrey /Shutterstock

Making steel — a material critical for construction and for the manufacturing of cars, buses and trucks, among other things — creates more carbon emissions than any other heavy industry, amounting to about 8 percent of the world’s total greenhouse gas emissions. Decarbonizing the steel industry doesn’t just require electric power or a clean way of producing high temperatures. It requires moving away from coal-blast furnace technology, which is how steelmakers currently perform the fundamental task of combining carbon (from the coal) with iron to create steel. 

Experts in green steel told Aventine that over the last several years the desire to decarbonize steel has gone from being merely an idea under discussion to a commitment shared by almost every industry player in the U.S. and Europe, to make steel carbon-free or to dramatically reduce carbon emissions from steel manufacturing by 2050. 

At the moment there are a few ways to achieve this goal, all extremely difficult and expensive; the technology furthest in the development pipeline and the most practical for existing steelmakers is a process called direct reduced iron, which uses green hydrogen as the initial fuel source. For this technology to work, steelmakers require both the production of green hydrogen for the direct reduced iron process and electric arc furnaces to transform the iron into steel. For the first step, green hydrogen is created by splitting water molecules into their component parts — hydrogen and oxygen — using only renewable energy, a process that isn’t new but has been historically so expensive to achieve — and so taxing on power grids — that it has been used only in very limited applications. For the second step, in order for the green hydrogen to be used to purify iron ore for steel, the steel plants must replace their coal-blast furnaces with electric arc furnaces, another significant and expensive investment. (There are some electric arc furnaces already in operation, but they are mostly fueled by natural gas.) While some generous tax credits are incentivizing the production of green hydrogen plants for the first time in the U.S., Europe and elsewhere, the creation of green hydrogen for steel production is going to drastically increase the price of steel, by up to 100 percent according to some estimates. 

Despite the cost increases, clean energy and manufacturing tax incentives have proven enough to spur initial production, even if speculation lingers about the customers for such expensive steel. Sweden's H2 Green Steel is building the largest green hydrogen-powered plant in Europe, expected to become operational in 2026, while other green hydrogen-focused projects are underway in China and other parts of Europe. In the U.S., the Biden administration announced in March grants worth about $500 million each for steelmakers SSAB and Cleveland-Cliffs to build the first facilities that could accommodate green hydrogen, mainly by updating already existing steel plants. 

Aventine spoke with five different experts in both the industry and in policymaking to understand what it will take to decarbonize steel.

Players battle for the ball as a corner kick is taken during a Premier League match at Bramall Lane, Sheffield. Laurence Griffiths/Alamy

Liverpool, U.K. To the uninitiated, the scramble of soccer players in front of the goal during a corner kick can look like chaos. To anyone who knows the game, it’s a carefully orchestrated exercise. The Economist reports that AI is now firmly in the second camp, ever since DeepMind developed TacticAI, a so-called soccer tactician, in collaboration with Liverpool Football Club. Trained on data from 9,693 corner kicks taken between 2020 and 2023 in the English Premier League, TacticAI — described in an article in Nature Communications — is able to analyze configurations of players on the field to predict who will first touch the ball and whether a goal might be scored. It is also able to suggest potential improvements to these set plays, which human experts were unable to distinguish from human-suggested plays and were preferred over the original plays 90 percent of the time. The big caveat here is that corner kicks are well-rehearsed; it will be much harder for any AI to predict or make proposals for how a team should play in the messy 90 minutes of a full game.

Brazil. Buzzing through the skies above the Brazilian states of Pernambuco and Paraná are a swarm of specially modified drones designed to release mosquitoes into the wild. But these are no ordinary mosquitoes: They are sterile male mosquitoes, intended to curb the reproduction rates of the insects and in turn slow the spread of mosquito-borne diseases such as dengue. MIT Technology Review reports that the drones, operated by the startup BirdView, can each carry as many as 17,000 mosquitoes and were carefully tested to ensure that they could transport the insects without damaging them. The team marks the insects with specific colors before release so that it can monitor how widely the insects spread. The difficulty may come in finding the funding to extend the trials to other geographies, given that the worst-affected locations for mosquito-borne illnesses tend to be some of the poorest, but BirdView hopes to train communities to use the technology for themselves in order to save on costs.

San Diego, U.S. Bobbing around in the San Diego Port is a little white and gray boat with a big claim to make: It’s the first all-electric tugboat in the U.S., Canary Media reports. Commissioned by the maritime logistics company Crowley — and named eWolf, for the company’s first-ever tug from the 1900s, Seawolf — it is designed to reduce emissions. It will also help curb nitrogen oxide levels around the port, which tugboats produce in disproportionate amounts. The boat itself is powered by a 6.2 megawatt-hour battery system and two electric motors, but its most intriguing element is a dockside charging system. This is actually a microgrid: two 1.5 MWh batteries covered with solar panels that help keep the batteries charged. (The microgrid is also connected to the local California grid, which it can both feed and get power from.) The tug moors at a shoreside station between operations to recharge, but thanks to the microgrid doesn’t tax local energy supplies while doing so. The microgrid also insulates the boat from being unable to recharge during power outages. It’s not clear exactly how much the boat cost, though Crowley told Canary Media that the boat and the charging setup “cost about twice as much as a diesel-powered tugboat of similar size.”

8 Google Employees Invented Modern AI. Here’s the Inside Story, from Wired
4,300 words

In 2017, eight Google researchers — one of whom had recently left the company — published an academic paper that would change the way the world thought about artificial intelligence. The paper in question, titled “Attention Is All You Need,” was the first articulation of a then-new AI approach that the team called transformers — the technology that now underpins large language models like OpenAI’s ChatGPT, Google’s Gemini and Anthropic’s Claude. This profile of the eight researchers goes deep on what the transformer technology is, how it was developed, and why Google didn’t move faster with a technique that had been developed on its own campus. It’s essential reading if you want to understand the history of today’s chat-based AIs.

How AI mathematicians might finally deliver human-level reasoning, from New Scientist and Can A.I. Make Plans? from The New Yorker
2,600 words and 4,000 words respectively

Ask most AI researchers what needs to be done to produce an AI with general intelligence and the list will be long. Two of the more important skills that artificial intelligence still lacks are reasoning and planning, and these articles each take a look at techniques being employed to make advances in those areas. One of the most interesting parts of the research described here is the way these experimental models are being trained: Some algorithms are studying mathematical proofs so that they can achieve human-like levels of reasoning; others are examining the complex strategy game Diplomacy so that they can learn to think ahead. While the progress so far is incremental, these two stories make it clear that researchers are not shying away from trying to imbue artificial intelligence with capabilities that are still considered uniquely human. 

The battle for the internet’s visual future, from Digital Frontier
2,300 words

A long story about file types may not sound like particularly inspiring reading. But this essay takes a close look at the battle between two new image types, each vying to be the dominant way of displaying high-fidelity, small file-size pictures online, and it reveals the very human tensions and power dynamics around new technologies. In one corner is JPEG XL, an evolution of the JPEG file type that’s existed since the ’80s; in the other is AVIF, based on one of the algorithms used to compress video so that it can be more easily sent across the internet. JPEG XL is backed by a community of developers and image experts; AVIF is backed by a group of large technology companies that includes Google and Apple. And Google, in a mic-drop moment two years ago, made the decision to remove support for JPEG XL from its Chrome browser. Battle lines are now forming and are messy at best: Apple is part of the consortium behind AVIF but is also backing JPEG XL, for instance. It’s far from clear which format will win out, but there seems to be plenty of drama left in the story.