Author: a16z New Media; Translator: Deep Tide TechFlow
This week's a16z chart report covers four topics, each worthy of a separate article: the Jevons effect triggered by declining AI costs, the true scale of tech giants' capital expenditures, Kalshi's prediction market is beating professional forecasting agencies, and the widespread postponement of life milestones for Americans aged 30. The data sources are solid, and the perspective is calm and restrained, making it a high-quality reference for understanding the intersection of current technological and macroeconomic trends.

DExit… Is it a real trend or an illusion?
... Delaware remains the top choice for business registration in the United States, but this position is quietly shifting: According to Ramp data, Delaware's share of new company registrations has been declining since 2023, with a drop of approximately 10% in the third quarter of 2025. History doesn't simply repeat itself, but it often rhymes…maybe. Delaware wasn't always a haven for business registration. About a century ago, Delaware replaced New Jersey—the former "mother of trusts"—as the top choice for business registration. New Jersey lost its edge because then-Governor Woodrow Wilson's attempts to curb "corporate abuses" significantly worsened the state's business environment. Delaware's corporate law, modeled after pre-Wilson New Jersey law, naturally welcomed departing companies and, together with the Delaware Court of Chancery, spent nearly 100 years building a reputation as a mature and impartial forum for resolving disputes between corporations and investors. However, what took a century to build began to crumble in just a few years. Right or wrong, the Delaware Court of Chancery has adopted a more lenient stance in shareholder litigation in recent years (especially in several high-profile cases, including but not limited to Tesla), and companies have begun to genuinely relocate their places of registration. Goodnight, and best of luck, Delaware. This is at least the mainstream narrative, but other data suggests a more complex picture. First, even the founding myth of Delaware itself is not entirely accurate. It wasn't until the 1980s (about 60 years after Governor Wilson took office) that Delaware truly surpassed New Jersey to become the state with the most registered businesses in the United States: New Jersey's reign lasted much longer than the mainstream narrative suggests. The catalyst for Delaware's eventual overtaking was likely the passage of a series of laws related to director responsibilities, which made it particularly attractive to publicly traded companies, coupled with the self-reinforcing network effect that created its own momentum. Secondly, regardless of what's happening with high-profile publicly traded companies (and the companies in the Ramp data), Delaware as a whole still looks to be performing well, or even better: According to data released by the Harvard Law School Corporate Governance Forum, Delaware's share of the total number of businesses in the U.S. actually increased significantly from the end of 2024 to 2025. In fact, if you're looking for a clear "DExit" case, it's probably this one, and it has nothing to do with Tesla, but rather involves a specific company structure: Wyoming LLCs began to grow rapidly around 2015. Why? This is likely related to specific asset protection and privacy provisions in Wyoming LLC law, which the state itself promotes as a "cowboy cocktail." In short, the point here isn't that DExit isn't happening (because at least some data suggests it is—even if it's just a few high-profile companies moving out, it's significant), but the reality is certainly more complex than the mainstream narrative portrays. The reality is that Delaware still enjoys the advantage of being the default option, not to mention all the network effects tied to it, which are difficult to shake. We previously published an earlier version of this chart, but as more data has been added, the results have become increasingly striking. Token costs are decreasing, while token consumption is increasing: Since the beginning of this year, the price of paid tokens has dropped from approximately 90 cents per million tokens to 50 cents, 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. This is encouraging. Remember when someone said that once newer and better GPUs are released, older GPUs become unwanted? This doesn't seem to be the case: According to Silicon Data, rental prices for NVIDIA H100 and A100 have both increased this year. The market shows no signs of oversupply of computing power; in fact, it seems that even the surface of existing demand hasn't been reached. This comparison isn't a perfect analogy, but if history can provide a reference, it may still take some time before we truly understand what an "AI-driven" economy looks like: From Faraday and Henry's initial discussions of electricity to the real explosion of industrial productivity in the first half of the 20th century, approximately 100 years passed. While the technological iteration cycle has indeed accelerated since the 1820s, the variables involved in a platform-level shift remain extremely numerous. Roy Amara famously said, "We always overestimate what will happen in the next two years and underestimate what will happen in the next ten years." Capital expenditure, viewed on a coordinate axis, reveals this set of timeless data: AI capital expenditure is massive. Please see the following comparison: AI capital expenditures in 2026 are projected to be close to the total net new loans issued by U.S. banks in 2025: Capital expenditures are approximately 33% higher than total U.S. corporate income tax revenue and about three times the total tariffs: Capital expenditures are approximately six times the total military budget of any non-U.S. G7 member country: So, yes, capital expenditures are indeed very large. Kalshi enters the field of macro forecasting. Federal Reserve researchers believe their market forecasting is quite good. At least on one metric, Kalshi's forecasts of the federal funds rate have outperformed professional forecasting agencies: For the federal funds rate forecast 150 days later (i.e., after three FOMC meetings), Kalshi's mean absolute error is very close to that of professional forecasting agencies. However, unlike surveys that only provide modal path snapshots every six weeks, Kalshi provides a continuously updated complete probability distribution… We found that Kalshi's median and mode forecasts have a perfect record of accuracy the day before the FOMC meeting, which is a statistically significant improvement compared to federal funds futures forecasts. In other words, while all forecasters start at roughly the same level, Kalshi's "continuously updated" predictions optimize over time, ultimately achieving a "perfect prediction record" the day before the official interest rate announcement. Furthermore, Kalshi outperforms predictions in the futures market. Kalshi's advantages extend beyond the federal funds rate. As Federal Reserve researchers have pointed out, since there are no other options markets for macroeconomic indicators such as inflation, growth, and unemployment, Kalshi is the only place that can provide a "high-frequency, continuously updated, and probability-rich benchmark" to reflect the "public's" judgment on the direction of these economic indicators. This sounds quite important. The Delay of Adulthood This is a thought-provoking chart, with (a few) comments: The percentage of people in their 30s who achieve major life milestones has been declining quite steeply since at least the 1980s. Fewer and fewer people in their 30s are: living independently; having been married; living with children; or owning their own home. The only exception is college enrollment—the proportion of 30-year-olds with a bachelor's degree has nearly doubled since 1995. So, is going to college worth it? A milestone? More like a millstone grinding your neck, right? Perhaps it is, perhaps it isn't, but a sense of "buyer's regret" seems to be pervasive.