NVIDIA sells the chips that companies like Anthropic and Open AI use. Now it is spending almost $13 billion to acquire Hugging Face, one of the most important platforms for open AI models. Is NVDIA trying to prop up the competition against OpenAI and Anthropic? The economics are more compelling than just a foreclosure story. In this episode, we examine two forms of vertical integration through the lens of a simplified AI supply chain, consisting of **chips → model development → developer toolkit**. OpenAI and Anthropic package models with their developer tools. Instead, NVIDIA and Hugging Face would combine the chip layer and a common toolkit but keep the model layer open for competition. One telling clue we also look at: NVIDIA is paying about 86 times Hugging Face’s reported annualized revenue vs. about 15 times the annualized revenue for Cursor. That implies that NVIDIA might be buying something that Hugging Face doesn’t directly monetize: a stronger open-model ecosystem that could drive more demand for NVIDIA compute. This leads to a counterintuitive possibility. NVIDIA may want OpenAI and Anthropic to succeed—but it may not want to depend on them.

Read the full article and graphic analysis: https://blog.econworks.com/p/why-nvidia-wants-openai-to-have-competitors?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

The problem with a cartel is that its agreement is illegal. If a member of a cartel cheats, the other members can’t simply sue in court and get a judge to enforce the agreement. So cartels need their own systems of monitoring, detection, and punishment.

On this episode of How Cartels Really Work, we take a look at the economics of cartel discipline through the alleged shipping-container “Box Cartel,” which allegedly used production quotas, financial penalties, and 87 cameras across 49 production lines to ensure compliance.

What is surprising is the punishment. If a cartel member cheats, the ultimate response may be a return to competition and the loss of future cartel profits. Then AI changes the narrative. AI could watch the competition, see any deviations, and respond faster than human managers. Even an artificial intelligence can be economically retaliated against.

But prison can't. That presents a new question for antitrust enforcement: If the decision-maker is an AI, who is the law trying to deter?

Read the full article and graphic analysis: https://blog.econworks.com/p/cartels-episode-4-how-do-you-enforce?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

Sysco is planning to buy Jetro Restaurant Depot. But this is not a normal horizontal merger. Sysco delivers food to restaurants, mainly. Restaurant Depot operates cash-and-carry warehouse-based stores. Those channels may serve very different needs for large restaurant chains.

Both, however, could be good choices for small, independent restaurants. That raises a simple antitrust question: do the two channels actually check each other? Maybe the answer lies in Sysco’s own experiment in Houston. Prior to the Jetro acquisition, Sysco had opened cash-and-carry stores called Sysco To Go to serve smaller foodservice customers.

If Restaurant Depot were to respond to that, it would create competition in cash-and-carry. If the prices Sysco delivered also moved, then it would indicate competition across channels. In this episode, we consider why restaurant size, local competition, and Sysco’s experiment in Houston may be more important than national market shares in understanding the proposed merger.

Read the full article and graphic analysis: https://blog.econworks.com/p/when-delivery-meets-cash-and-carry?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

How can economists find a cartel if there is no agreement? Nasdaq stocks were quoted in eighths of a dollar in the early 1990s. But economists discovered that for many heavily traded stocks, half of those potential price increments were rarely used. This pattern became known as the Nasdaq “missing eighth.”

The research was made public, the dealers acted quickly, and the spreads tightened. Later investigation unearthed evidence that reached far beyond the initial statistical anomaly.

In Episode 3 of How Cartels Really Work, we take this episode to discuss a fundamental distinction in antitrust economics: empirical evidence can detect suspicious behavior, but an unusual pattern is not proof of an agreement on its own.

Sometimes the economist discovers the pool. Sometimes the economist finds a reason to start looking.

Read the full article and graphic analysis: https://blog.econworks.com/p/cartels-episode-3-can-data-reveal?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

The proposed acquisition of Jetro Restaurant Depot by Sysco seems very different from the failed merger between Sysco and US Foods. Sysco mainly ships food to restaurants. Restaurant Depot operates on a cash-and-carry basis, meaning the restaurant owners come and pick up the food themselves.

But different channels don’t necessarily mean different competition. Delivery may be a must for a large restaurant or national chain. But for a smaller independent restaurant, the choice may be between paying for delivery or doing some of the logistics itself. That means the overlap is critical for smaller restaurants. And Houston might provide a particularly useful test.

Before purchasing Restaurant Depot, Sysco opened its own cash-and-carry stores, called Sysco To Go, in Houston. That provides economists the opportunity to ask what happened when Sysco entered the pickup channel.

Did prices at nearby Restaurant Depot respond? Did delivered prices from Sysco change? If both did, then the boundaries between delivery and cash-and-carry may be less significant than they first appear. This episode explores what the FTC would need to prove if it wanted to challenge the Sysco–Jetro deal—and why the best evidence may come from the actions of restaurant owners rather than from national market shares.

Read the full article and graphic analysis: https://blog.econworks.com/p/when-delivery-meets-cash-and-carry?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

In the early 1990s, Nasdaq stocks were quoted in eighths of a dollar. $20.00, $20.125, $20.25, $20.375, etc.

But economists William Christie and Paul Schultz found an odd pattern. For many actively traded stocks, market makers mostly used even eighths. Instead, prices acted as if quotes were changing in quarter-dollar increments.

Why? Because?

The empirical evidence alone cannot prove that dealers agreed to keep spreads wider. But once the research was released, odd-eighth quotes started to show up, and spreads tightened considerably.

Then government investigators dug deeper, finding evidence that went far beyond the original statistical pattern.

In this episode of How Cartels Really Work, we examine the case of the Nasdaq to make a key distinction:

Economics can identify the clue without requiring the agreement to be established.

Sometimes the data doesn't tell investigators what happened. They point investigators in the right direction.

Read the full article and graphic analysis: https://blog.econworks.com/p/cartels-episode-3-can-data-reveal?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

To summarize, AI might soon grant a single researcher access to much of the capacity that used to belong to an entire academic team: theory, coding, empirical analysis, data work, writing, and project management. That opens a more profound question: what if every economist owns a research factory? This episode is about more than the simple idea that AI will help academics write more papers.

Research factories are about efficiency, not output maximization. They could enable researchers to explore ideas sooner, to kill weak projects sooner, to explore strong ideas more fully, and to concentrate on the areas of research that are still rare. We also revisit Brooks's law. More AI agents and analyzes don't automatically improve a project.

Someone still has to coordinate the factory and decide which work is worth doing. If artificial intelligence can democratize the technical research capabilities in the future, then the advantage may shift back to the things that are harder to automate: the research question, evidence, institutional knowledge, judgment, and ground truth.

Read the full article and graphic analysis: https://blog.econworks.com/p/what-happens-when-every-economist?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

Picture two gas stations across the street from each other. One increases its price. The other one follows. And it happens over and over again. The pattern looks funny. But does it prove price-fixing?

In episode 2 of How Cartels Really Work, we explain why it is a lot harder to prove a cartel than it is to describe one.

We follow six observations that make the hypothetical case incrementally more compelling: simultaneous price increases, repeated matching, prices rising independently of costs, communications between competitors, increasing specificity of discussions, and finally evidence of an agreement. The problem is that competitors can behave in similar ways for reasons unrelated to an illegal agreement. Companies spy on one another. They react to the same market conditions.

And in concentrated markets, the best course for a firm may depend on what it expects its rivals to do. So the important question is not whether prices moved in the same direction. It is the cause they moved together. This episode discusses the difference between parallel conduct and agreement, direct and circumstantial evidence, and why economists looking for suspected cartels often start with pricing patterns, not end with them.

Read the full article and graphic analysis: https://blog.econworks.com/p/cartels-episode-2-why-price-fixing?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

What if every economist had an AI research factory? Academic research has often been a team effort, with complementary skills: theory, empirical work, coding, data management, writing, and project management.

Many of those capabilities are increasingly available to a single researcher on demand via AI. But it doesn’t mean the aim is to produce as many papers as possible.

In this episode, we explore how AI might change the organization of academic research, why Brooks’s Law remains true inside an AI research factory, and why efficiency should be measured by the quality of the work that survives—not just the quantity of research produced.

And once all economists have the same technical capabilities, the competitive advantage may shift to better questions, better evidence, institutional knowledge, judgment, identification, and ground truth. The research factory could be an infrastructure. The quality of what we decide to build with it will still matter.

Read the full article and graphic analysis: https://blog.econworks.com/p/what-happens-when-every-economist?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe

Two gas stations simultaneously raise their prices. And then they do it again. Their prices have not changed much. Their owners have been even talking. Is that enough evidence of price-fixing?

In episode 2 of How Cartels Really Work, we take a look at one of the key issues in cartel enforcement: the line between suspicious parallel behavior and evidence of an agreement.

Starting with a simple example of a gas station, we move through six observations of increasing importance. This includes simultaneous price increases and repeated matching, as well as competitor communications and evidence of an actual agreement.

We explore how the same pricing pattern can arise from two very different stories: either the competitors colluded to raise prices or they observed and reacted to each other independently.

The important question is not whether prices moved together. That’s why they moved together.

Read the full article and graphic analysis: https://blog.econworks.com/p/cartels-episode-2-why-price-fixing?r=562wri

Explore more visual economics content: https://econworks.com

YouTube: https://www.youtube.com/@EconWorks-d3e

Substack: https://blog.econworks.com

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit blog.econworks.com/subscribe