Hey! Thanks for being here for Edition #16. You're one of 260+ subscribers, and I genuinely appreciate every single one of you.
An M&A firm in the UAE recently made me an offer. They'd own my brand, my network, my business, every deal I originate and the execution process. In exchange: $9k a month.
Saying no was easy. I get 500k+ impressions a month and originate dozens of deals every week across real estate, PE, secondaries and pre-IPO. My brand and network get me headhunted, bring in referrals, originate deals and attract capital. I built both from scratch.
I'll rent my brand all day long - sponsorships, partnerships, collabs. But I won't sell it. At least not for $9k a month.
Funnily enough, personal brand came up again on this week's podcast - more on that in Sponsor Tips.
Let's get into it.
Nvidia is now underwriting AI data centres. Last week it bought another $1.5B of SB Energy, SoftBank's data centre and power developer, just ahead of its Nasdaq IPO - doubling its total stake to $3B. The shares were priced at 90% of the IPO price and are non-voting, so SoftBank keeps control. SB Energy is targeting a $50B valuation and a $5–7B raise, off H1 2026 numbers of $138.7M revenue and a $3.2B net loss.
The bigger story was filed in August. Nvidia has guaranteed up to $105B on the leases at SB Energy's Ohio campus - a former US uranium enrichment plant - where the tenant is OpenAI. That's 4.25GW of initial IT load, an option on another 3.8GW, a 20-year lease, Nvidia as exclusive chip supplier and 10GW+ of new power generation planned. So Nvidia backs the landlord, OpenAI rents the building, and the building runs on Nvidia chips. The $105B is a cap, not cash - Nvidia only pays if the leases fail and the site can't be re-let or sold. But that guarantee is what makes the project bankable. Chips aren't the bottleneck anymore. Power is. So the chip company is now underwriting the power.
Also this week:
Grab buys the underwriting, not the customers. Grab is paying $1.49B in cash for 60% of Atome, one of Southeast Asia's largest BNPL and consumer lending platforms - its biggest cash acquisition to date. $260M goes straight into Atome as growth capital. Atome brings a $1B+ loan book, $470M of 2025 revenue (up 80%) and AI credit scoring built for thin-file borrowers; Grab brings 54M monthly users and the data on how they ride, order and pay. The last 40% is priced at 13x EBITDA / 2.5x revenue with a $2B floor and $4.5B cap, paid roughly two years after closing - so if Atome underperforms, Grab pays less. Target: a $6B+ loan book and $500M financial services EBITDA by 2028.
A $29B succession sale becomes a new family office. Sysco has agreed to buy Jetro Restaurant Depot at a $29.1B enterprise value - $21.6B in cash plus $7.5B in Sysco shares, leaving the family with 16% of Sysco. It's the biggest deal in Sysco's history. Nathan Kirsh, now 94, started with a £1,200 inheritance and a corn mill in Swaziland in 1958, lost most of it in the 1980s and rebuilt in Brooklyn. His kids don't run the business, so the family is reportedly building a New York single-family office to manage the proceeds - hiring investment chiefs, a full IC and a team to oversee outside managers. Most families sell and hand it to a private bank. The Kirsh family is building the bank. GPs will be lining up.
Is the AI race now a race for electricity?
What's crossing my desk and my conversations this week:
A man who talks to family offices all day. I sat down with Niccolò M. Mottola, Director at the Global Family Office Investment Summit, for Episode 3 of the Gamma Prime podcast. He works directly with family office principals across Europe, the US, the UAE and Asia. From where he sits, 40% of the deals he sees are data centres and energy - not AI itself, but the infrastructure behind it - and green-powered DCs are the strongest of the lot. The other sector standing out is longevity, with a lot of deal flow coming out of the Nordics. It fits how families think: in decades, not fund cycles. The one that surprised me was a fund raising $100M+ to tackle loneliness.
Watch the full episode: https://www.youtube.com/watch?v=cabd0goc-uk
Bali's first dedicated leasehold villa lender. Bali pulls in billions from international buyers every year, but banks won't touch most of it because foreign buyers usually hold leasehold, not freehold. So after 10+ years in the market, this founder put $10M+ of his own money into building the lender he couldn't find. Loans go up to 50% of property value, in USD through a Singapore entity, at 11–15% to borrowers on 2–3-year terms (up to 7), with 3–4 deals a year. Backers get 10% a year, paid monthly, senior secured, no management or subscription fees. His line: in Bali you pay cash or you don't buy at all. He's betting that changes.
A Hong Kong family office selling pre-IPO blocks. Their LPs are first-generation mainland entrepreneurs - some now run their own funds - so they sell blocks on behalf of their LPs as well as buying for them. 2–3 LP meetings a week, only 3–5 names at a time, no LOI or proof of funds, data room after NDA, a direct call with the GP and $5M minimum tickets. They've already been in Binance and MiniMax and are now hunting OpenAI, Blue Origin and Fireworks. Right now they're placing a block of Moonshot AI ahead of its Hong Kong IPO, looking for buyers in the US and Middle East. Buying American AI, selling Chinese AI.
Are you buying or selling pre-IPO blocks?
Update: The UK brownfield AI data centre now has a signed customer.
I've covered this one before: a modular AI/HPC data centre campus on a brownfield site in the Warrington sub-region of North West England. The site is owned unencumbered by the sponsor, a UK infrastructure group with c.25 years in facilities management and mission-critical engineering. Demolition is underway, and grid power is already on site.
When I last wrote about it, the big gap was offtake. There were no signed contracts - just advanced talks and non-binding LOIs. That's changed.
The sponsor now has an executed Order with an NVIDIA Cloud Partner, dated 10 September 2026, under a Master Services Agreement. It covers 16MW over 60 months, with a stated total contract value of £172.8M (ex-VAT) - roughly £180k per MW per month. Fitting, given the Nvidia story at the top of this edition: the chip ecosystem isn't just supplying these sites, it's renting them.
Demand doesn't stop there. The same client is asking for more capacity, and several other NVIDIA Cloud Partners are at LOI stage. The roadmap now runs to 410MW of additional capacity: a further 60MW at this site (15MW in 2027, 45MW in 2028) plus seven new 50MW sites across the North West, phased from April 2027 to January 2029. That's 426MW in total if it's all delivered - a big if, and the sponsor says so: everything beyond the first 16MW is still subject to definitive contracts, funding, power and site readiness.
The original ask was a £65.5M structured debt facility for Phase 1 - 8% interest, a 2% management fee and 15% of five-year profit, with an estimated c.£115M back to lenders (c.12% p.a.). Security is a first charge over the land-owning SPV, a debenture, share pledge and step-in rights, but no charge over the land itself unless the ticket is £100M+. The capital conversation has now widened to growth equity, project finance, equipment finance and JVs to fund the expansion.
If you're a lender or infrastructure investor looking at UK data centres, reply to this email to set up a call.
Why cold outreach rarely works with family offices - and what does.
This came straight out of my conversation with Niccolò, and it lines up with what I see every week.
Family offices are the fastest LPs if they like an opportunity - and the person presenting it. But they're small teams, usually 5–20 people, not a PE firm with 100 analysts. So they do due diligence on you before they do it on the deal. They want to know who you are, your background and what value you bring. Values matter. First impressions matter.
That's why cold outreach struggles. Principals don't publish their numbers, and if you reach a gatekeeper, their job is to not pass you through. Niccolò came from sales - his cold email success rate was around 7%. What flipped his business was LinkedIn: inbound meetings where the other side has already seen who you are, and now they're the ones selling you on the call.
His playbook:
Pick one niche and post daily. LinkedIn rewards consistency - put yourself in a box ("the family office guy") and it pushes you to the right people.
Headline and image do most of the work. A little exaggeration is fine. Untrue is not.
Referrals compound. Families co-invest constantly. Win one family's trust, and the next introduction comes from them.
Track record before size. Raising $2B with no website, no LinkedIn and no prior raise? That's a quick no. So is anything that looks dodgy - he won't touch gold bullion deals, because it's his reputation on the line.
Face-to-face still wins. Research the family before you meet - interests, sports, kids. In the UAE, the first hours of a meeting won't be about the deal at all.
His advice to his younger self: start building the brand ASAP.
Which is why I'm not selling mine for $9k a month.
A snapshot of what's currently moving through Capital Arbitrage:
Real Estate:
Hong Kong - Land Purchase & Rezone - $70M Debt/Equity
Netherlands - Office Portfolio Conversion - $150M Debt/Equity
Dubai - Luxury Villa Flip - $16M Debt/Equity
UK - Brownfield Data Centre - £65M Debt
USA - New York Multifamily - $2M Equity
Canada - GPU Financing - $89M Debt
Australia - Debt Purchase / Refinance - $350M
Secondaries:
Direct GP access to Blue Origin, Databricks, Stripe and Firehawk Aerospace.
This isn't the full pipeline - just a flavour of what's active. If you're an investor or lender active in any of these asset classes or geographies, reply to this email.
That's all for this week. See you next Friday.
— Jordon
P.S. I get 500,000+ monthly impressions on LinkedIn and a growing list of private capital readers right here. If you'd like to get your company, fund, or raise in front of this audience, just reply to this email.