They sell the fix for the thing that broke Ashburn twelve days ago.

S: "I needed a Massachusetts installer for Raynora's channel. Their president skipped my term sheet and sent back a pre-executed 17-page agreement on their paper — classic anchor."

T: "Negotiate a counterparty who'd already shown he negotiates hard, without killing the relationship."

A: "I built a locked one-page addendum through ten versions plus a four-column discussion guide, traded rather than demanded — deleted chargebacks to win payment timing and a trailing 12-month liability cap, took their $2.40 TPO rate as the explicit trade for $2.05 commercial — and dropped two of my own asks live when he correctly called them overreach."

R: "We signed. The two-document playbook became my template for four other counterparties. The lesson your JD names exactly: knowing when to push and when to concede is the same skill."

S: "The installer I most wanted — Empower, rebranding to Bold Energy, 29-state footprint — sent an 18-page agreement with two $75,000 liquidated-damages clauses, uncapped liability, one-way indemnity, and a clawback that triggered when their lender pulled funding for any reason."

T: "Decide whether the deal I wanted most could be papered at a risk level I could survive — and hold that line against my own enthusiasm."

A: "I rebuilt the risk allocation in a 16-section addendum with order-of-precedence: LDs deleted, liability capped at trailing 90-day commissions, clawback scoped to my own misrepresentation, tiered 7/14/30-day change notices, mutual indemnity. That addendum was my written floor before emotion could move it."

R: "[FILL — one true sentence on the end]. Walking from the deal you want most is the real test. Decide the walk-away line before you fall in love with the deal."

S: "Mid-negotiation on a channel agreement — terms substantially settled, including a disclosed 10% margin cap and a four-step commission formula."

T: "Before signature, verify the upstream installer network actually existed where they claimed."

A: "Real-time diligence found no East Coast operating presence — registered-agent addresses instead of warehouses, no Google Business footprint, no response by phone, email, or text. I pulled the agreement and sent a due-diligence email asking specifically who installs in the DMV, Maryland, New Jersey, and Massachusetts."

R: "The deal never closed, and that was the win. The cheapest place to catch counterparty risk is before your signature — I learned that the expensive way watching an installer I depended on fail."

S: "I needed a residential solar financing product that cleared both the IRS true-lease test and Islamic finance constraints at the same time — two rulebooks that usually fight each other."

T: "Design terms where every clause had a number behind it, not just legal comfort."

A: "I locked a 20-year Ijara structure with an 8% nominal target IRR on the payment stream, Year-6 purchase-option gate past the §48E recapture window, FMV methodology floored at 5%, and benchmarked clause-by-clause against four national contracts including hell-or-high-water payment language. Separately I built a §48E beginning-of-construction reference covering safe harbor and transferability."

R: "A financeable product with nine mandatory lessor obligations, including a production guarantee. When your JD says modeling is teachable — I taught myself the version that matters: what a term does to the economics."

S: "Across Boundless, Team Sunshine, Empower, Convert, and Yellowlite I was re-solving the same negotiation from scratch every time — and losing time to document hygiene and self-negotiation."

T: "Turn judgment into a repeatable kit so the first draft is aggressive-but-reasonable and the live call has a shared map."

A: "I standardized on two artifacts: a locked executable addendum as PDF, plus an editable four-column discussion guide — Section / Proposed language / Rationale / Their notes. Doctrine: never self-negotiate v1; let them redline. Column labels matter — 'Rationale' not 'Why,' 'agreed' not 'counter.' I also wrote Raynora's own 24-section Sales Partner Agreement — the paper others negotiate against."

R: "Your JD's last bullet — develop playbooks, templates, frameworks — isn't aspirational for me. I have one. It's priced in kilowatts instead of gigawatts."

"Limitation of liability is the clause that decides whether a bad quarter is survivable or existential — especially when the customer's downside is a killed training run on billions of GPU capex. My position is mutual consequential-damages waiver, an aggregate cap tied to fees paid, and a negotiated carve-out list — not an open-ended one. And I watch whether indemnity sits outside the cap, because that quietly makes the cap meaningless.

In my own paper: Team Sunshine — I added a trailing 12-month liability cap and a six-year survival sunset to their indemnity rather than fighting the clause head-on. Empower — their original had no cap at all, including reputational damages; my addendum capped at the greater of actual damages or trailing 90-day commissions. Same surface as your JD, smaller zeros — the zeros are the teachable part."

"Yes — and I removed them. Empower's agreement had two $75,000 LD clauses plus a $100-per-item materials return. I treated that as a risk that needed to go to zero on my book, not get trimmed: deleted all three in the addendum and limited remedies to actual damages under the liability cap. At ON scale I'd expect delay and performance LDs to stay in the deal — so the job becomes capping them, making them sole-and-exclusive for that failure, and tying delay LDs to the module actually late, not the whole campus. I haven't papered that at hyperscaler scale; I have papered the judgment call of whether an LD belongs at all."

"First question: is it mutual, and is it proportional to fault? Second: does it sit inside or outside the liability cap? Third: is the scope tied to something insurable?

I've caught asymmetric indemnity twice — Convert and Empower — where my side covered negligence and theirs only gross negligence, and in one case their investors sat inside my obligation. I pushed mutual and got investors out. On ON's product I'd also expect IP indemnity to cut both ways given the patent position, and fire/thermal exposure to be sized against the Lockton programme, not left as prose."

"Payment terms are where I engineer cash-flow risk out of the deal rather than hoping collection works. In my archive: 30/40/30 milestone splits, 45-day post-PTO backstops, 14-day payment deadlines, redline locked at submission, proportional tranches within seven business days for cash and commercial. On Team Sunshine I split timing by product — install-completion for TPO/lease, seven-business-day proportional for cash and commercial. On Empower I added a 21-day post-installation backstop regardless of when they got paid by their financing partners.

Security-wise I've negotiated non-suspendable portal access surviving disputes until the last submitted deal is paid, anti-set-off language, and a forgivable signing advance with a partial-recovery floor. At ON the same logic scales up to LCs, parent guarantees, and milestone front-loading against cell and transformer procurement — the working-capital gap your Lombard Odier facility exists to cover."

"Every marketed number becomes a warranty or a performance guarantee someone has to stand behind. Ride-through capability becomes an availability or ZVRT commitment — measured how, over what window, with what exclusions. Efficiency and capacity retention become LD triggers if you miss them. Modular 3.2 MW blocks become the unit for delay damages — late modules, not a late campus. If I see a datasheet claim, my first commercial question is: who warrants it, for how long, what's the remedy hierarchy, and is it back-to-back with the OEM."

"First I ask if we can engineer it out — restructure the deal so the risk can't happen. If not, cap it — bound the exposure in time and dollars. If someone else holds it cheaper, transfer it — insurance, warranty, indemnity. Whatever's left, price it into the commercial terms. And if it can't be priced, that's what walking away is for.

Every clause is that question wearing different clothes. I'd rather bring executives a residual-risk statement — 'indemnity capped at 12 months trailing, six-year sunset' — than a problem labeled 'uncapped indemnity.'"

"I'd lead with the economics, not the redline — what's the expected return this term is protecting. Then residual risk after mitigation, named on the ladder. Then whether it fits our risk appetite — some risks are cheap for ON to hold because you own the hardware and the O&M team. Then whether it fits strategy — a below-market Crusoe-scale deal can be right; a rich deal that locks a bad precedent can be wrong. Recommendation first, residual risk stated, ladder rung named. That's the JD line about advising executives on how terms affect project economics, made operational."

S/T: "Counterparty held a hard line on TPO pricing at $2.40/W; I needed economics that worked on commercial."

A: "I didn't fight the TPO number in isolation — I priced the concession. Accepted $2.40 on TPO as the explicit trade for $2.05 on commercial, and paired it with payment-timing and liability-cap wins elsewhere so the package cleared my floor."

R: "Signed deal. A concession with a price tag is pricing, not losing — and it's the difference between being the department of no and being the commercial counterweight."

"Unpriceable exposure I can't cap, transfer, or get paid for — uncapped consequential damages on a compute-outage scenario with no mutual waiver and no path to insurance. Unverifiable counterparty — if diligence can't confirm who performs, I pull before signature; I've done that. Precedent that poisons the book — a one-off concession Sales wants that becomes the template for the next ten campuses. And anything that breaks bankability: if lenders won't finance against the term, the 'win' isn't a win for a company that owns assets.

I've walked or pulled three times in ninety days — Empower on liability architecture, Convert on three named dealbreakers, Aegis when the installer network didn't check out. The floor was written down before the call so enthusiasm couldn't move it."

"I name the risk in dollars and residual terms, not vibes — then offer the ladder options: engineer it out, cap it, transfer it, price it, or walk. I want Sales to leave with a path to yes at a risk level Finance can live with, not a veto. If the strategic value is real — Crusoe-scale — I escalate with the residual risk stated clearly so the executive decision is informed, not ambushed. Holding a line without becoming obstruction is the job."

"I'd expect to sit under Finance — the role is filed there, and deal summaries go to the CFO for executive review, so my working assumption is Andrea Petersen. Day to day I'd partner with Alberto Avila, VP Finance for Data Centers, and Kevin Rooney on structured finance — he came from Live Oak, which has been on ON's lender side, so he knows how credit reads these contracts.

The tension the JD names is Sales: Eric McGaha as VP Sales is the counterweight relationship — bookings versus margin and risk. Procurement under Lauren Wong matters for back-to-back warranty and OEM terms. And I wouldn't wait for Legal — you're hiring Contracts Counsel in parallel, which is exactly why the posting says you don't wait for Legal to tell you a clause is a problem. I'd want you to confirm who this seat reports to."

"Alan Cooper, CEO and co-founder — he's led the major financings personally; Norman Cooper chairs the board. Ricardo de Azevedo, CTO and co-founder — AI UPS architect, ex-Goldman and Barclays, wrote 'Don't Call Me BESS,' which is the product thesis in one essay. Andrea Petersen, CFO — treasury and reporting at public-company standards; that's the line this role likely serves. Dax Kepshire, COO — hyperscale delivery, ~20 years scaling storage companies. David Fernandes, CIO and founding partner — ex-CFO, asset deployment and project financing.

What that map tells me commercially: product vision sits with Ricardo, capital and risk discipline with Andrea, and delivery with Dax — so a commercial-negotiations seat under Finance is the bridge between a bookings machine and a balance sheet that still has to clear lenders."

"On the Crusoe side, Chris Dolan — Chief Data Center Officer — is the name on the customer posture for the 5 GW AI UPS program. On delivery, Clint Summers at Rosendin — senior director of power engineering — shows up as a named EPC/power partner in your materials. And on your side of the table for capital, Pathward and BridgePeak are on the $77.6M credit; Lombard Odier on the development facility. I don't pretend I know those people personally — but I know which desks the contracts have to survive."

"Because the deliverable is margin and risk, not bookings and not document administration. Sales — McGaha's world — originates; Legal is being built and can't be the bottleneck; Finance — Petersen's world — owns whether the term is financeable and whether the residual risk fits appetite. The JD literally says this isn't a contracts-administration role and that you're the commercial counterweight to Sales. That's an operator seat with a CFO audience."

"You started around 2015–16 — Cooper and de Azevedo — as a storage developer and integrator out of Miami into Latin America. You became a US IPP with real ERCOT positions, raised a Series B with Ultra Capital and Phalanx in 2023 against hundreds of megawatt-hours there, then stacked development capital with Lombard Odier and a $77.6M Pathward–BridgePeak construction credit — materials put you past $250M raised.

The interesting move is 2026: February AI UPS launch — patented inline medium-voltage system, NLR-validated — and in July the Crusoe 5 GW program across hyperscale campuses for 2026–27. What strikes me commercially is you're three businesses at once — manufacturer, integrator, asset owner, historically retaining most of what you build — so contract terms don't just allocate risk, they decide whether a project gets financed. That's a more interesting problem than vendor contracting. And it's why a Finance-owned commercial seat opened now."

"June 2023 — $20M Series B, Ultra Capital and Phalanx, tied to ~300 MWh in ERCOT. April 2024 — $25M Lombard Odier development facility. January 2025 — $77.6M Pathward construction credit with BridgePeak; company materials say $250M+ raised to date, plus tax-equity and ITC-transfer capacity on the order of $120M. The commercial read: you've been building a balance sheet that can own assets, not only sell equipment — so every customer contract has a second reader, the lender's credit committee. I'd want to know which terms those credit partners treat as non-negotiable, because those are the floor of my position."

"Reston puts you next to Data Center Alley and PJM — the densest concentration of the load problem your product solves. Greenhouse shows ~56 open roles across Reston/Tysons, Houston, Miami, Atlanta, Amarillo, Mexico City, Argentina — including Contracts Counsel and a Director of Sales for Data Center & Critical Power in parallel with this seat. The trigger for opening commercial negotiations now is almost certainly Crusoe: multi-campus, multi-year, change orders and payment security at a scale where margin lives or dies in the paper. LATAM isn't legacy either — José Manuel Díaz Pérez as President LATAM, and my interviewer on Bogotá time, tells you that geography is still live."

"July 21, 2026 — ON and Crusoe announced up to 5 GW of AI UPS across multiple hyperscale campuses, phased 2026–27. Chris Dolan is Crusoe's Chief Data Center Officer on that posture. Same Crusoe that developed and operates Abilene for Stargate — so this isn't an abstract logo; it's one degree from the biggest AI build-out in the country. Commercially that means concentration risk and commissioning risk over years — which is exactly why payment security, change-order mechanics as chip and cooling architectures move, and liability architecture aren't boilerplate. I'd rather price those correctly once than renegotiate them under a live campus schedule."

"Every megaproject made a different power bet, and each bet creates a different contract surface. Stargate / Abilene — OpenAI, Oracle, SoftBank; Crusoe operating the flagship — behind-the-meter self-supply to dodge interconnection queues. Meta Hyperion in Louisiana — customer pays for the grid: Entergy gas plants, renewables, transmission, BESS, ratepayer fights at the PSC. xAI Colossus in Memphis — mostly off-grid gas, 168 Megapacks smoothing GPU swings — parallel storage solving the problem ON argues needs an inline system. AWS Rainier for Anthropic — grid-dependent, maximum ride-through exposure. Microsoft Fairwater — the bear case: Atlanta reportedly with no UPS and no diesel; NOGRR-style mandates are the counterweight. ON's thesis is regulation is about to punish the naked-grid bets — and Ashburn on July 22 was the PJM proof point."

"Twelve days before this conversation, a transmission fault in Data Center Alley knocked roughly 3 GW of load off PJM in seconds. I live about twenty minutes from that epicenter. ON's entire product thesis is that computational load has to ride through those events instead of protecting itself by vanishing — which is also what destabilizes the grid. July 2024 was the first warning shot; July 2026 made it impossible to treat as theoretical. If I'm pricing contracts for AI UPS, that's the failure mode in the warranty, the LD, and the availability definition."

"ERCOT moved first: NOGRR 282 — ride-through for large computational loads at or above 75 MW — effective August 1, 2026, which is essentially now. FERC has been clarifying large-load and co-located interconnection in PJM. The commercial implication is a regulatory-change mechanism in multi-year agreements: if the rule tightens mid-build, that's a change order, not silent absorption. TIEC has challenged ERCOT's authority — so regulatory risk itself has to be allocated, not wished away."