Refuse to guarantee Oct 5 for Core42 if CIQ is deploying Ubuntu - give a likely range unless the customer demands a guarantee

September 11, 2026 at 3:04 PMstrategyhigh

Situation

Asked by Adam Jackson in #core42-build-and-deploy whether CIQ could commit to an Oct 5 completion starting Sept 18, Peter answered no for the full-Ubuntu-deployment scope: there is absolutely a chance CIQ would hit it, but CIQ cannot guarantee it. He then separated the scopes - if it is just Fuzzball, and Ubuntu is already stood up on the racks and networked, and CIQ gets access Sept 18, then absolutely CIQ could hit that date. He gave the rule for how to answer the customer: be VERY careful about guarantees in a new environment and an OS that is not ours, because a guarantee puts CIQ in a position to lose future business when this is the first time CIQ is doing this; if Core42 MUST have a guarantee that is one thing, but if they only want a likely range then give the range and do not make an unforced error. He then asked the four scoping questions that decide which scope applies - is Ubuntu already installed, is networking already set up, is there a provisioning machine available for ELM, and exactly how many machines.

Reasoning

The date is answerable only once the scope is pinned, so Peter refused to answer the question as asked and split it. The asymmetry drives the rest: hitting an unguaranteed date costs nothing, missing a guaranteed one costs the next deal, and CIQ has never done this deployment before so its own estimate carries real error bars. Volunteering a guarantee nobody asked for is therefore pure downside - the unforced error. The four questions are the mechanism for converting the open-ended risk into the bounded scope CIQ can commit to.

Additional Context

Reprises the Aug 31 Core42 decision to hold the commitment at five weeks while signalling two-to-three as likely. The Sept 10 Core42 Pre-Work Followup meeting confirmed the shape Peter had pushed for - access around Sept 18, three-week build, live by Oct 6 - and Nathan carried the same discipline into the SOW, writing it so FIPS is not required but Core42 pays for an Ubuntu One subscription if they insist.

Observed Evidence

Direct quotes giving the refusal, the conditional yes, the general rule about guarantees, and the four scoping questions - all in the same channel within hours.

Matching Patterns

45%
Decide on the Precedent, Not the Case(names the precedent risk - losing future business - as the reason for the answer, answers the rule rather than the specific date request)
40%
Reclassify to Route - Decide Who Owns It, Then Give a Criterion Not a Verdict(refuses the question as framed and identifies what it actually turns on - scope, supplies a criterion for the customer conversation rather than a date)
30%
Protect Engineering Capacity(refuses a commitment that would bind engineering to an unbounded scope)

Confidence Breakdown

35/35
Evidence
28/30
Pattern
20/20
Source
12/15
Corroboration

Reasoning Depth Analysis

Org Signal:Tells the deal team that Engineering will give honest ranges and will not be talked into guarantees to make a deal feel safer, and tells engineers that Peter will absorb the discomfort of saying no to a date rather than pass the risk down to them.
Who Affected:Adam Jackson and Bjorn, who have to sell a range rather than a date. Ryan Smith and Ahmer, whose ghost deployment next week exists to shrink the error bar. The Shinns, who are the ones who would have to deliver an Oct 5 guarantee.
Precedent:Establishes that a first-of-its-kind delivery gets a range, and that a guarantee is something the customer must actually demand rather than something CIQ volunteers to look confident.
Consequences:Real: the SOW was written without a FIPS requirement and with the cost of FIPS pushed to Core42, and the team scheduled a ghost deployment specifically to de-risk the date.
Timing:Now because the SOW was being finalized in the same window - once the date is in a signed document the option to give a range is gone.

Source

reflection

AI Confidence

95%

Related Context

💬
#core42-build-and-deploy

slack

I would be VERY careful about guarantees, in a new environment and an OS that is not ours. A guarantee puts us in a position to lose future business, when this would be the first time CIQ is doing this. So if we MUST give a guarantee date because Core42 demands one, that is one thing. But if they do not demand one and want to know likely range, we should give likely range and not make an unforced error here.

💬
#core42-build-and-deploy

slack

No. Starting Sept 18, we could not guarantee an Oct 5 completion date if we were deploying Ubuntu. There is absolutely a chance we would hit it. We cannot guarantee it. IF it is just fuzzball, and Ubuntu is already set up on those racks and networked together and we get access Sept 18 - then I would argue that absolutely we could hit that date.

💬
#core42-build-and-deploy

slack

The four questions: IS Ubuntu already set up on all machines in the rack or is CIQ doing that? Is networking already set up for all machines? And is there a machine set up that is available as a provisioning machine that we can run ELM on? Is it 600 nodes - exactly how many machines are we talking about?

🎥
Core42 Pre-Work Followup, Sept 10

fathom

Nathan Blackham: potentially we are going to be starting a week from tomorrow where we will get access, and we will have roughly three weeks, so the 5th of October... we should go do the deployment on it with test hardware and go work out the kinks next week.

Outcome

GB300 still undelivered past 10/31 with network down; no broken guarantee

Rating: 4/5

Decision ID: c1035137-3281-4df4-a08f-3cc3fec8bee7