Author: a16z New Media; Compiler: Shenchao TechFlow

This issue of a16z’s Chart Weekly covers four topics, each worthy of its own article: the Jevons effect triggered by falling AI costs, the true scale of capital expenditures by technology giants, the Kalshi prediction market is beating professional forecasting agencies, and the overall postponement of life milestones for people in their 30s in the United States. The data source is solid and the perspective is calm and restrained. It is a high-quality reference for understanding the intersection of current technology and macro trends.

DExit...is it a real trend or an illusion?

Delaware is still the first choice for US business registration, but this status is quietly loosening:

Delaware’s share of new company registrations has continued to decline since 2023, falling by about 10% in the third quarter of 2025, according to Ramp data.

History doesn't simply repeat itself, but it often rhymes... maybe.

Delaware has not always been a mecca for business registration.

About a century ago, Delaware replaced New Jersey—the original "Mother of Trusts"—as the top place to register a business. New Jersey lost its advantage because then-Governor Woodrow Wilson tried to curb "corporate abuses," which greatly deteriorated New Jersey's business environment. Delaware's corporate law was modeled on the laws of New Jersey before the Wilson era, so it was naturally willing to welcome outgoing companies. Since then, it has worked with the Delaware Court of Chancery and spent nearly 100 years establishing a reputation as a mature and fair place to resolve disputes between companies and investors.

However, what took a century to build began to falter in just a few years. Rightly or wrongly, the Delaware Court of Chancery has taken a more lenient stance on shareholder lawsuits in recent years (notably several high-profile cases, including but not limited to Tesla), and companies have begun to actually move their domiciles elsewhere. Good night and good luck, Delaware.

That's at least the dominant narrative, but other data suggests the picture is more complicated.

First, even the Delaware founding myth itself is not entirely accurate.

It was not until the 1980s (about 60 years after Governor Wilson took office) that Delaware truly surpassed New Jersey and became the state with the largest number of business registrations in the United States:

New Jersey’s dominance lasted far longer than the dominant narrative suggests. The catalyst for Delaware's eventual overtaking is likely to be its passage of a series of laws related to directors' liability, which are particularly favored by listed companies. In addition, the network effect continues to strengthen itself, forming its own inertia.

Second, regardless of what is happening to high-profile public companies (and the companies in the Ramp data), Delaware overall still looks like it is doing well, if not more than well:

According to data released by the Harvard Law School Corporate Governance Forum, Delaware's share of the total number of U.S. businesses has actually increased significantly from the end of 2024 to 2025.

In fact, if you want to find a clear case of "DExit", it's probably this, and it has nothing to do with Tesla, but involves a specific corporate form:

LLCs in Wyoming began to grow rapidly around 2015.

Why? This is likely related to the specific asset protection and privacy provisions in Wyoming's LLC laws, with the state itself promoting this corporate structure as a "cowboy cocktail."

In short, the point here is not to say that DExit is not happening (because at least some of the data suggests that it is indeed happening - even if it is just a few high-profile companies moving out, it is still significant), but the actual situation is certainly more complicated than the mainstream narrative presents.

The reality is that Delaware still enjoys the advantage of being the default option, not to mention all the network effects tied to it, and those are hard to shake.

We posted an early version of this graph before, but as we added more data, the effect became even more striking.

Token cost decreases and Token consumption increases:

Since the beginning of this year, paid token pricing has dropped from approximately 90 cents to 50 cents per million tokens, while the number of tokens processed has almost doubled, from approximately 6,000 to 12,000.

This is a classic Jevons effect. The cheaper AI becomes, the more AI we use. Delightful.

Remember when people said that when newer and better GPUs come on the market, no one wants the old GPUs?

This does not seem to be the case:

According to Silicon Data, rental prices for both the Nvidia H100 and A100 have increased this year.

The market is far from showing signs of an oversupply of computing power. On the contrary, it seems that the surface of existing demand has not even been scratched.

This comparison is not a perfect analogy, but if history is any guide, it may be a while before we can truly see what an "AI-driven" economy will look like:

It took about 100 years from the time Faraday and Henry first began discussing electric current to the real outbreak of the wave of industrial productivity in the first half of the 20th century.

Technology iteration cycles have indeed accelerated since the 1820s, but the variables involved in a platform-level shift are still extremely numerous.

Roy Amara has a famous saying: "We always overestimate the changes in the next two years, but underestimate the changes in the next ten years."

Capital expenditure, view it on the axis

Look at this set of data that will never go out of date: AI capital expenditures are huge.

Please see the following comparison:

AI capital expenditures in 2026 are expected to be close to the total net new loans of U.S. banks in 2025:

Capital expenditures are approximately 33% higher than total U.S. corporate income tax revenue and approximately 3 times the total tariffs:

Capital expenditure is approximately 6 times the total military budget of any non-U.S. G7 member state:

So, yes, the capex is really huge.

Kalshi enters macro forecast

Federal Reserve researchers think prediction markets are pretty good.

On at least one indicator, Kalshi's forecast performance on the federal funds rate has exceeded that of professional forecasting agencies:

For the federal funds rate forecast 150 days out (that is, after 3 FOMC meetings), Kalshi's mean absolute error is very close to that of professional forecasters. But unlike surveys that only provide a snapshot of the modal path every six weeks, Kalshi provides a continuously updated full probability distribution... We find that Kalshi's median and mode forecasts have a perfect forecast record the day before the FOMC meeting, which is a statistically significant improvement over the fed funds futures forecast.

In other words, although all forecasters have a similar starting point, Kalshi's "continuously updated" forecasts will continue to be optimized over time, eventually reaching a "perfect forecast record" one day before the interest rate is officially announced. Additionally, Kalshi has outperformed futures market forecasts.

Kalshi’s advantages don’t end with the federal funds rate. As Fed researchers pointed out, since there are no other options markets for macro indicators such as inflation, growth, and unemployment, Kalshi is the only place that can provide "high-frequency, continuously updated, probability distribution-rich benchmarks" to reflect the "public's" judgment on the direction of these economic indicators.

Sounds pretty important.

Delay of adulthood

This is a thought-provoking chart, with (a few) comments:

The proportion of 30-year-olds achieving major life milestones has been declining quite steeply since at least the 1980s.

Among the 30-year-olds, fewer and fewer people:

Living independently;

Once married;

Living with children;

Own your own home.

The only exception is college enrollment - the share of 30-year-olds with a bachelor's degree has nearly doubled since 1995.

So, is college worth it?

Milestone? More like a stone grinder for your neck, right? !

Maybe so, maybe not, but "buyer's remorse" seems to be in the air.