Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
83 77 88 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
69 49 87 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
69 49 87
All types combined: 81 (likely 76 to 86, high confidence), from 87 claims: 77 factual claims, 5 predictions, 5 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Sam Altman's claims
“This is hugely complex and expensive and we are still doing this in like a sort of bespoke one-off way, although it’s getting better.”
Sam Altman, OpenAI CEO 9 August 2025 Transcript: Sam Altman Interview on Huge Conversations Podcast (Singju Post)
OpenAI builds AI compute infrastructure (chips, memory, networking, data centers, energy) in a bespoke one-off way, though this is improving.
Mostly true True in substance, wrong in a detail.
The second judge would not resolve this claim; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judge as of August 2025. True: reporting or company statements from around that time show that OpenAI's compute buildouts (for example Stargate sites, cloud and chip deals, power arrangements) were negotiated, designed or built individually, project by project, rather than from a standardized, repeatable template, with some signs of the process becoming more systematic. Mostly true: the buildouts were largely custom, but there is little evidence either way that the process was improving, or some parts (for example chip procurement) were already standardized. Mixed: about half the main parts were custom and half were standardized. Mostly false: OpenAI was mainly using standard designs or off-the-shelf cloud capacity, with only isolated bespoke elements. False: evidence shows OpenAI's infrastructure was built in a standardized, repeatable way at that time.
Evidence
- Announcing The Stargate Project OpenAI · 21 January 2025 Own record
“The buildout is currently underway, starting in Texas, and we are evaluating potential sites across the country for more campuses as we finalize definitive agreements.”
- Announcing The Stargate Project OpenAI · 21 January 2025 Own record
“As part of Stargate, Oracle, NVIDIA, and OpenAI will closely collaborate to build and operate this computing system.”
- Oracle and OpenAI Contracts: Deal Size and Backlog Luminix · 15 October 2026
“Additional Stargate capacity | OpenAI / Oracle | 4.5 GW additional; with Abilene, over 5 GW under development, “over 2 million chips”; jobs estimate over 100,000 | Announced July 22, 2025”
- Building the compute infrastructure for the Intelligence Age OpenAI · 29 April 2026 Own record
“These projects are complex, and they require the right combination of power, land, permitting, transmission, workforce, community support, and partner readiness.”
- Oracle and OpenAI Contracts: Deal Size and Backlog Luminix · 15 October 2026
“Stargate project announcement | SoftBank, OpenAI, Oracle, MGX (equity); Arm, Microsoft, NVIDIA as technology partners | Intended $500 billion of investment over four years, $100 billion to be deployed immediately; 10 GW goal | Announced Jan. 21, 2025; four-year horizon”
- Oracle and OpenAI Contracts: Deal Size and Backlog Luminix · 15 October 2026
“Additional Stargate capacity | OpenAI / Oracle | 4.5 GW additional; with Abilene, over 5 GW under development, “over 2 million chips”; jobs estimate over 100,000 | Announced July 22, 2025; no payment start date in this post”
- Oracle and OpenAI Contracts: Deal Size and Backlog Luminix · 15 October 2026
“The only Oracle filing that sizes the economics is the June 30, 2025 8-K. Safra Catz said Oracle had signed multiple large cloud services agreements, “including one that is expected to contribute more than $30 billion in annual revenue starting in FY28,” and warned that amounts ultimately recognized “may vary.” The customer was not named.”
- Stargate Advances with 4.5 GW Partnership with Oracle OpenAI · 22 July 2025 Own record
“With SoftBank, we’re moving quickly on site assessments and reimagining how data centers are designed to power advanced AI.”
- OpenAI's Stargate Data Centers Are Taking Longer and Costing More Than Its Competitors' Distilled (newsletter by a reporter who authored Cleanview's behind-the-meter data center report) · 17 June 2026
“In December 2024, the company ordered 10 GE Vernova turbines, and then in June 2025, it bought another 19 units that could collectively generate more than a gigawatt of power.”
- Stargate Data Centers Site-by-Site: Status, Dependencies, and What to Watch Measured AI (Substack) · 31 March 2026
“The Abilene reversal is not an isolated event. It is a signal that the Stargate buildout is unfolding not as the coordinated mega-project presented at the White House, but as a collection of individual sites at sharply different stages of development — each with its own developer, power procurement approach, financing structure, and execution risk.”
- Stargate Data Centers Site-by-Site: Status, Dependencies, and What to Watch Measured AI (Substack) · 31 March 2026
“Beyond its energy position, SB Energy’s construction approach relies on modular data center units manufactured at SoftBank’s Lordstown, Ohio facility — a former Foxconn EV plant acquired for $375 million in August 2025 and being converted into modular equipment production.”
Reasoning
An independent site-by-site review (E10) describes the Stargate buildout as a set of individual sites, each with its own developer, power procurement approach, financing structure and execution risk, rather than one coordinated template. Other sources point to deal-by-deal arrangements: separate turbine purchases (E9) and a separate 4.5 GW Oracle capacity announcement (E6, E8); Oracle's 8-K also disclosed a large cloud agreement with an unnamed customer (E7). There is only weak evidence that the process was becoming more systematic by August 2025. OpenAI's own talk of "reimagining" data center design (E8) and SB Energy's modular units (E11) point that way; the Lordstown facility for those units was acquired in August 2025 and was still being converted to modular production when the later article was written. So the bespoke part holds up, while the "getting better" part is only thinly supported.
Cite this claim
Sam Altman, “OpenAI builds AI compute infrastructure (chips, memory, networking, data centers, energy) in a bespoke one-off way, though this is improving.” (said 9 August 2025). Verdict: mostly true. Reliability API, claim c_0c34de9ccac1, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_0c34de9ccac1
Claims without a verdict
Is a verdict here wrong, or do you speak for Sam Altman? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.