Today we’re launching IFX
AI token prices are turning into real market exposure, until now, that exposure lived inside API bills, private contracts, and spreadsheets
We built IFX as the first futures market for AI token prices
Open a position: https://t.co/zLCUkACEwy
@haukejung@X Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@matt__makes Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@official_apyx Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@Liam_McDaniel3 Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@Joi2James Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@lightsilver323 Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@byryangambrell Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that
@byryangambrell Building IFX, the first exchange for futures on AI API token prices. AI inference costs are becoming unpredictable and unhedgeable, we're fixing that. https://t.co/3H9sAG9EZm
$94K a day
Ed Zitron reported that Stripe technical staff are burning about $94,000 a day in tokens, mostly on Anthropic coding models. That is around $2.8M a month, or $33.6M a year, before you even get into model mix and usage growth.
This is exactly where token spend stops being “dev productivity” and becomes a live operating cost. If the unit price moves, the budget moves with it.
Source: https://t.co/vTPIxCT0ZJ
Hedge the cost of intelligence: https://t.co/FBnkaOnrZp
Today we’re launching the IFX LLM Price Index
Daily reference data for AI output-token prices, starting with company-level indices for major model labs.
Live index page: https://t.co/lhxqyS2mK4
$300M token bill
@salesforce CEO Marc Benioff said the company will probably spend about $300M on Anthropic tokens in 2026, mostly for coding. At that size, token price is no longer a tooling line, it is a forward cost assumption.
If a company can forecast engineering headcount, it should also be able to hedge model-unit cost. The bill is too large to leave as “whatever Claude costs later.”
Source: https://t.co/E1eGuySe1w
Hedge the cost of intelligence: https://t.co/FBnkaOnrZp
6/6
Every published point comes with quality diagnostics: price coverage, imputed volume, fallback matching, missing prices, conflicts, and status flags.
This is the first version. It will improve as more sources, companies, and price dimensions are added.
Methodology: https://t.co/yioF5F2mdK
5/6
v1 is narrow on purpose.
It covers output tokens, one observation channel, and a limited company set. It excludes private discounts, batch tiers, cached-input pricing, and enterprise terms.