A research paper at the U.S. Federal Reserve praised the usefulness of prediction markets — specifically looking at Kalshi — in getting a real-time handle on economic policy.

“Kalshi’s forecasts for the federal funds rate and [the U.S. Consumer Price Index] provide statistically significant improvements over fed funds futures and professional forecasters, all while providing continuously updated full distributions rather than infrequent point estimates,” according to the paper published on Thursday.

And the markets, in which retail investors can buy contracts in virtually any yes-no question in such diverse fields as economics, politics and sports, are looking at topics on a live basis that other sources of information don’t.

Prediction markets “provide unique insights — particularly for variables like [gross domestic product] growth, core inflation, unemployment and payrolls, for which no other market-based distributions currently exist.”

And in this study, Kalshi predictions “perfectly matched the realized federal funds rate by the day of each meeting since 2022, a feat not achieved by either surveys or futures.”

Part of the secret sauce that sets prediction markets apart as a useful tool may be the inclusion of retail participants, which makes them “distinct from institutionally dominated markets,” the paper noted.

Public Bitcoin miners are planning about 30 gigawatts of new power capacity aimed at artificial intelligence workloads, nearly three times the 11 GW they currently have online, as they race to offset shrinking mining margins and reposition for the next growth cycle.

The buildout, compiled by TheEnergyMag across 14 publicly traded Bitcoin (BTC) miners, underscores how aggressively the industry is pivoting away from traditional hashpower amid persistently weak hashprice conditions.

On paper, the planned expansion amounts to what TheEnergyMag described as “a small country’s worth of power infrastructure.” In reality, much of the 30 GW sits in development pipelines, interconnection queues or early-stage plans, rather than operational facilities.

The widening gap suggests competition is shifting from ASIC efficiency to securing power, financing and delivering data centers on time.

“This is the megawatt arms race of the AI boom,” TheEnergyMag said, adding that monetization ultimately depends on whether AI demand remains strong enough to justify the scale of investment.

Related: The real ‘supercycle’ isn’t crypto, it’s AI infrastructure: Analyst

AI pivot delivers early revenue gains for some miners

The shift toward artificial intelligence infrastructure reflects an increasingly hybrid strategy among established Bitcoin miners, with some companies already reporting meaningful revenue contributions from AI and high-performance computing (HPC) workloads.

One example is HIVE Digital, which recently posted record quarterly revenue driven in part by its AI and HPC business lines. The company reported fourth-quarter sales of $93.1 million, up 219% year on year, even as Bitcoin prices declined during the period.

Investors, too, are attuned to the shift. Earlier this week, Starboard Value went public with its suggestion to Riot Platforms management that they accelerate the miner’s expansion into HPC and AI data centers.

The push to diversify comes as mining profits have taken a hit since the 2024 Bitcoin halving, which cut block rewards and squeezed margins across the industry.

Conditions have gotten even tougher since the fourth quarter, when heavy selling pressure sent Bitcoin tumbling from its record high above $126,000. Prices eventually stabilized in February, after briefly falling to below $60,000.

Despite these headwinds, US-based miners showed resilience at the start of the year, with output rebounding after a severe winter storm temporarily disrupted operations.

Related: Paradigm reframes Bitcoin mining as grid asset, not energy drain