Begin forwarded message:
From: Jared Birchall <j ared@excession.com>
Date: September 1, 2018 at 12:21:32AM PDT
To: Sam Teller @~12acex.com>
Cc: Shivon Zilis @.§12acex.com>, Reyna Ortiz @.§12acex.com>
Subject: Re: AI decisions
Thanks for this. I tend to agree with Sam's thought of maintaining a seat and converting what has already been
given into equity. As an investor, E would at least be able to keep tabs on progress, as well as keep the door
open for greater influence if desired.
One correction to Basic Fact #2, Elon's last donation was in August. He donates about $300k p/mo. But true
that his last multi million dollar donation was in 2017.
I really question what weight Disclaimers #3 and #4 really carry. In both cases an investor can choose to think
or act differently with no apparent consequence. It'll be interesting to see what E thinks.
I'm digesting the rest, but with funding it appears that the non-profit donations roll into the $500MM first round
1:1.
I'll be back. ..
On Fri, Aug 31, 2018 at 11 :33 PM, Sam Teller @.§12acex.com> wrote:
Thank you. Very helpful. Let's discuss with Elon next week. My instinct says it's better for him to maintain a
seat at the table, since even if Google is way ahead, OpenAI is (I believe) the clear #2 at this point.
On Aug 31, 2018, at 11:15 PM, Shivon Zilis @.§12acex.com> wrote:
Hey guys,
Realize there's a lot on the go, but I think it's critical to ensure Elon is able to come to the right decision for
himself on OpenAI now. Once they close $500M raise in this structure it is going to be a big deal and
relatively irreversible.
This is meant as a comprehensive doc to throw all critical things in one place so we can decide how to help
him, but I'll note that I don't have a lot of experience with financial documents so please correct any and all
things.
CONFIDENTIAL SPX-006742
Basic Facts
1. Elon does not currently have structural control since resigning from the board (did not opt for a proxy or
any other persisting governance).
2. He is the largest sole donor to date unless something has recently changed, though his last donation was
in 2017.
3. Altman clearly has a deep desire to work with Elon and have him aligned, but says he realizes that may
not be possible and respects Elon's choices. Greg and Ilya do not share those desires but seem cool with
either scenario and are leaving it to Altman.
4. Elon needs to decide a) if I how he wants to be affiliated b) what edits to financial structure he'd advocate
for if he's involved c) how he wants his donations to date to be treated.
5. A call with Reid seems like a good option in all cases since that's by far the more information gathering
and verifying conversation he can have.
Critical Elements of the Term Sheet
Disclaimers:
1. High risk
2. Potential for complete loss of capital
3. "Wise to view" in the snirit of a donation
4. Money may be meaningless in anAGI world
5. Ability to reinvest any and all cash flow into R&D
Tripartite Structure:
1. OpenAI Inc (non-profit)
2. OpenAI LP (investment vehicle)
3. OpenAI Holdings (employee equity vehicle)
Funding:
0. Initialization Capital Rolls over from non-profit($?? at I00x target - this is confusing)
l. Initial Round: $500M@ l00x target ($SOB, "Group 1")
2. Employee Pool: In holding company, worth $1 00B at target ($1 00B minimum, increasable at GP
discretion)
3. Second Raise: $10B+@ 15x target ($150B, "Group 2")
4. Possible subsequent raises at TBD
Roundup: We're talking about a target floor of $300B++(.!J.). of aggregate theoretical value. I realize money
is meaningless in AGI world, I only flag this to illustrate the potential gravity of how people will perceive
this entity.
Order of Distributions:
1. 100% to "Group l" until l x return
2. 25% to "Group I" and Employees until l00x reached, 75% to "Group 2" until lx
3. 100% to "Group 2 until 15x reached
4. Any excess back to non-profit
Other:
Timeline: Nothing in first 10 years except if GP advocates for earlier? (Jared would be curious to get your
read on exactly what they are saying ... I'm having trouble parsing yearly floating rates vs. actual target
payback)
Revenue: GP can reinvest revenue or distribute surplus revenue to investors
Fiduciary Duty: "fiduciary duties of the Nonprofit Board of Directors flow exclusively to the Nonprofit, not
the limited partners"
Strategic Investment: Can accept strategic investment if it doesn't "clearly subvert the redemption waterfall"
nor clearly detract from the advancement of our mission.
CONFIDENTIAL SPX-006743
Possible Outcomes:
Let it go entirely
If he chooses this I may recommend he ask for a name change away from "OpenAI". That's so linked to his
name and he came up with it. Since this is a huge transition from their current structure a rebranding doesn't
seem ludicrous.
Negotiate for a different structure
There are several things about this structure that may not pass muster with Elon, but I'm curious if he's
willing to engage with it. It is very weird, but will note that the goal here is to find a mechanism of getting
the billions needed for building AGI in the least harmful way, since there don't seem to be any harm-free
ways. Donations at that order of magnitude have proved impossible (Hoffman mentioned his frustration
with Gates on that front today), traditional for profit doesn't seem right, ICO was insane ... so??? Seems
worth at least thinking on what that structure is if not this one? Elon's answer may still be the only
passable capital source is "at Tesla" in which case they'd have to agree to disagree.
Participate in the current structure
If Elon does participate it seems like he'd need to regain some element of control, since the economic
element of this is rather meaningless practically if executed correctly and doesn't give investors any control
rights. The fundamental constraint here is they will not allow him to have 51 % control, so is it worth him
advocating for a seat at the table?
Next Steps
l . Have Elon glance at the term sheet.
2. Talk to Reid on fairly short order.
3. TBD based on Reid outcomes.
------You are welcome to stop reading now unless your name is Jared------
Trees jhr the jhrest (detailed questions that may or may not matter depending on ij'Elon wants to engage):
Term Sheet
1. *Heavy* control lies with the GP. Almost everything in the document is at the sole discretion of the GP.
Is the GP the OpenAI non-profit board? If so, what is that going to look like over time? You have to trust
those people a ton to buy into this if what you cared about what the good of infinity in the face of infinite
wealth.
2. "Surplus" revenue can be returned to investors at the discretion of the GP, and that capital would not
count towards the target amount. This is mildly scary because it may create a situation of consistent
pressure on the GP to create and disburse revenue.
3. What is this "corporate blocker" concept in the revenue section?
4. Initialization Capital... this is money rolling over from the non-profit but what is that and how much? I
thought they were roughly out of funds so it's a bit confusing. Also a bit confusing why the non-profit itself
gets an interest?
5. Employee interest ... if people are being distributed $1 00B in potential value will they be able to prevent
themselves from advocating to actually get it? I'm a bit confused about equity compensation of employees
if the goal is it's never meaningful. Maybe just pay people a hell of a lot?
6. Strategic investment. .. this cause feels extremely loosely written. May be worth a clarifying edit?
7. Was a little confused by Altman's comment of not taking equity now, but will take equity if it turns out
OpenAI is not building AGI but something else revenue generating?
8. What are the rough investment amounts and list of participants? Are Altman and Greg putting in
anything?
9. Is "wise to view in the spirit of a donation" sufficient for Elon?
CONFIDENTIAL SPX-006744
CONFIDENTIAL SPX-006745