Monday Night Links

• The trillion-dollar cluster—+4 OOMs from the GPT-4 cluster, the ~2030 training cluster on the current trend—will be a truly extraordinary effort. The 100GW of power it’ll require is equivalent to >20% of US electricity production; imagine not just a simple warehouse with GPUs, but hundreds of power plants. Perhaps it will take a national consortium. (Note that I think it’s pretty likely we’ll only need a ~$100B cluster, or less, for AGI. The $1T cluster might be what we’ll train and run superintelligence on, or what we’ll use for AGI if AGI is harder than expected. In any case, in a post-AGI world, having the most compute will probably still really matter.) [Situational Awareness]
• South Bow (SOBO) on May 29 announced open season results for its Prairie Connector project, securing 20-year binding commitments from Hardisty, Alberta to US delivery points. The company also revealed several key project details, giving the market more visibility on a potential new Canadian crude egress path. The project would add 380 km (~236 miles) of new 36-inch pipe and use 150 km (~93 miles) of preserved 36-inch pipe, plus two existing pump stations. Based on a typical oil pipeline flow velocity of 3-10 ft/sec, a 36-inch line implies ~300-1,100 Mb/d of theoretical capacity. SOBO is targeting a final investment decision (FID) by mid-2027. [East Daley Analytics]
• A year and a half ago, our city’s one bookshop went up for sale. My wife and I bought it. The place had 20,000 books, a good music system that probably played 3,000 hours of Bach per year, and a black cat named Raven. It even had an entire room just for theology and philosophy books. (Steubenville, Ohio, is an unusual town.) Now we’ve been running the shop for a little while. Experientia docet, said our ancestors. Experience teaches. I prefer Vivaldi to Bach, so you’ll hear more Vivaldi around here now. Raven died and my children are now tending a litter of six kittens, grooming a replacement. And experience has taught us something heartening: Our customers have great taste in books. I write this because I hope it will be as great a consolation to you as it has been for us. People tell us all the time that civilization is finished, that the world is coming to an end. But then we look at our sales details and we smile. [John Byron Kuhner]
  
• In the early 1970s, the Bretton Woods Agreement—the post-world war pact that instituted a fixed-exchange rate regime for the major world nations— had begun to show its structural flaw. Finance ministers were finding it increasingly difficult to dictate the value of currencies relative the dollar in a world where value changes were constant and information and capital free flowing. On August 15, 1971, President Nixon announced an emergency economic package that sent a seismic shock through the entire financial world. On that day, the United States suspended the dollar's convertibility into gold thereby ending fixed exchange rates between currencies. The Chicago Mercantile Exchange was the first major futures exchange to recognize the market potential of the upheavals unleashed by this event. Supported by Nobel laureate Milton Friedman, the CME was the first exchange to assert that the principles of agricultural commodities futures could be applied successfully to financial instruments. Thus, currency futures—which began trading on the CME's International Monetary Market (IMM) on May 16, 1972—ushered in the era of financial futures, thereby forever changing the scope and utility of futures markets. One year later, the Chicago Board of Trade launched the Chicago Board Options Exchange (CBOE) and added a new dimension to the repertoire of risk management instruments. Three years later, Treasury bond futures at the Chicago Board of Trade made their debut and became the most actively-traded financial instrument. Within a decade, a vibrant new industry was born that subsequently opened the curtain on the index markets of the 1980s. The successes of these markets propelled the futures and options industry to unparalleled greatness. In the last decade alone, the volume in U.S. futures and options skyrocketed from 76 million contracts in 1979 to a record of 323 million contracts in 1989. These successes also prompted University of Chicago Professor Merton H. Miller, 1990 Nobel laureate in Economics, to nominate financial futures as "the most significant financial innovation of the last twenty years." [Leo Melamed]
  
• The Silicon Data Token Expenditure Index, which tracks total spending on large language model usage, has roughly doubled since late 2025 even as the price of a single token has fallen more than 90% since 2023. This is Jevons paradox in action. As tokens get cheaper, companies don't spend less but instead run more AI agents, automate more workflows and generate more code, pushing aggregate expenditure higher even as the unit cost of intelligence collapses. [Apollo]
• From our vantage point at KKR, the current U.S. policy mix looks meaningfully different from the 2000-2019 period, which was defined by low growth, low inflation, global interdependence, and tighter fiscal constraints. In the current environment, we expect our asynchronous global recovery thesis, characterized by rolling recoveries and rolling recessions across regions and sectors, to not only continue but also gain broader acceptance. If we are right, the long-term implications for asset allocation could be significant, including the need to own more Real Assets across portfolios. Indeed, in a world where dispersion is rising and traditional diversification is less dependable, we continue to favor return streams anchored in hard assets and collateral-based cash flows. These exposures do not eliminate volatility, but they can improve the controllability of outcomes through seniority in the capital structure, stronger recovery prospects, and cash flows tied to essential activity in the real economy. [KKR]
• If this were true—that an LLM, lacking consciousness, must instead constantly confabulate its own behavior and motives and reasoning—then, even if an LLM is extremely intelligent, we should expect errors to accumulate differently in LLMs vs. humans. And as philosopher Toby Ord has demonstrated, this is exactly what’s observed. In his analysis of METR’s data on AI task-length doubling times, Ord identified that the available data also fit there being a simple “half-life” for the success of an AI agent, and so have a “constant hazard rate” for long tasks. As Ord writes about what this suggests concerning an AI’s prospects for completing long tasks: "the chance of failing at the next moment is independent of how far you’ve come—just like how the chance of a radioisotope decaying in the next minute is independent on how many minutes it has survived so far." Ord even shows that humans, when their success rate at long tasks is analyzed, appear to deviate from an equivalent constant hazard rate. A strong contender as to why is because we have interpretable access to our own previous thoughts and actions, i.e., our consciousness, in a way that LLMs simply don’t have. [The Intrinsic Perspective] 
• By choosing to treat the 1930s as a form of extreme payback for extreme excess—“no generation exempt”—Sorkin stages morality play rather than history. He also helps set policymakers up for the kind of grand theatrical action they are inclined to take anyhow whenever markets turn down. In other words, another 1933- or 2008-style rescue: flooding the market with liquidity, and stringing up wrongdoers and even the better Wall Streeters, such as the Mitchells whom Sorkin seeks to rehab. The same subpar results are likely to follow. [Amity Shlaes]
• In terms of AI progress per resource, or per dollar, things are probably getting worse on most measures. This is what the pessimism about scaling laws is getting at. Measures of quality are increasing far slower than the exponentially mounting costs. So why were people like Ilya Sutskever and Dario Amodei so impressed by the scaling laws? The answer is that there was a lot of headroom in compute — a lot of room to scale those costs.  Even if the resources needed would be thousands of times larger than the largest ever ML experiment, they saw that this (1) was still within the feasible set of things a large company could fund and (2) would still be a good deal given the vast potential benefits. For on the high end of possibilities, we are talking about something that could potentially replace more than half of all human labour in the world, and perhaps bring forward scientific and technological advances that human labour would have taken centuries to reach. In other words, while the costs could escalate wildly, a deep-pocketed project might reach benefits of extreme value before it ran out of money. And those benefits might be worth more than enough to justify those (very high) costs. More prosaically, we can note that it is entirely possible that financial returns will scale very quickly as a function of the technical measures of AI quality. If so, then even though standard measures of AI quality scale poorly as a function of resources, the financial returns might still scale very well as a function of resources. Indeed, if they scale better than linearly, that would create a paradigm of increasing marginal returns which would explain a landscape with a small number of players, each investing as much as they possibly can. [Toby Ord]
• It will come as a surprise to most that a company so stiff and humorless as CME Group currently is, once used to run an aggressive and somewhat jingoistic ad campaign in the mid 1970s for the purpose of celebrating free markets.  As I recently obtained a second set of these somewhat rare prints from a retired former CME marketing employee, it's worth taking a deeper look into the story behind them. [Trading Pit Blog]
• Mr. Chairman, there are no commodity exchanges in Moscow; there is no Peking Duck Exchange in China; there is no Havana Cigar Exchange. The farmers of those countries have no need for a mechanism that offers risk transference, price projection, or price protection. In those countries, the governments establish the prices at which farmers can sell their products. Consequently, the farmers' primary risk is entirely removed. Alas, by removing the risk, that system also removes the incentive. The sorry history of such systems is that they have been abysmal failures. [Leo Melamed]