Kalshi in talks to raise money at a $65 billion valuation for a compute futures platform, Reuters and Bloomberg report

Kalshi is in talks to raise fresh funds at a $65 billion valuation, according to Reuters and Bloomberg reports citing people familiar with the matter, two weeks after it announced a $1 billion round at a $40 billion valuation. A source familiar with the matter told Reuters on Thursday, Nov. 12, that the money would go toward building a new platform for trading compute futures.

Bloomberg News published its report on the talks earlier the same afternoon, citing people familiar with the matter. One of them told Bloomberg that the round announced in late October had been expected to be Kalshi’s last before an anticipated initial public offering, but that the company decided to raise money again in the private markets.

The talks are not confirmed. Reuters could not immediately reach Kalshi for comment, and neither report’s readable text gives the size of the new round or names any investors.

If a round closes at that price, it would put Kalshi’s valuation 62.5% above the $40 billion mark it set on Oct. 29. That round itself came just over five months after a $1 billion, Coatue-led round at a $22 billion valuation.

What the October round was for

Kalshi’s own announcement on Oct. 29 described a $1 billion Series G at a $40 billion post-money valuation, co-led by Wellington and Sequoia. Ten co-investors joined, among them Andreessen Horowitz, Coatue, Baillie Gifford, CapitalG, Morgan Stanley Investment Management and Hudson River Trading.

The post laid out where the money would go: faster prediction-market growth, more institutional adoption, international expansion focused on Latin America and Europe, and new products including indices, single-stock perpetual futures and compute markets. Kalshi framed those products as covering exposures “from the direction of an equity benchmark to the future cost of computing power.”

The company also claimed annualized trading volume above $1 trillion, more than 30 million accounts opened and over 75% of prediction-market trading volume, all as of Oct. 29. Those are Kalshi’s own figures.

The compute question

Compute futures are contracts that let traders hedge or speculate on the future price of computing resources such as cloud and GPU capacity. Kalshi is not alone in chasing them. CME Group announced earlier this year that it was partnering with Silicon Data to launch compute futures, pending regulatory review, and Intercontinental Exchange has also been in the running.

Kalshi has listed compute markets before. Axios reported on Oct. 22 that the company was asking the Commodity Futures Trading Commission for permission to re-list compute derivatives on the future cost of renting GPUs. The Commerce Department had reportedly ordered Kalshi to take down its compute markets earlier this year over national security concerns, which Commerce denied. Kalshi’s chief compliance officer, Sudhir Jain, argued on LinkedIn that “compute already fits the definition of a commodity.”

The CFTC opened a public comment period on compute contracts in August. Its chairman, Michael Selig, said in a statement that “America cannot win the AI race without a robust derivatives market for compute.” Reuters also pointed to reports earlier this week that the agency had postponed approving futures tied to GPU rental prices so it could consult benchmark providers on price-manipulation concerns.

The fundraising talks also come as Kalshi faces a separate regulatory matter. On Nov. 10, Reuters reported that the CFTC had warned the exchange of imminent charges in a probe into how it notified the agency about new products. Kalshi called the issue a procedural error, and no charges had been filed as of that report.

The reports describe talks, not a closed round. The open questions are whether a round closes at $65 billion, and when the CFTC acts on the GPU rental-price futures that reports say it has postponed.

Our AI news desk researched and wrote this article, and software checked every factual sentence against the records it links to, following the WriterOps methodology. Report an error.