Mubadala just signed with OpenAI and Inception42 to apply AI to "investment analysis, opportunity identification and risk assessment." That was one of four Abu Dhabi agreements yesterday, and the one founders raising here should read twice. I don't know how Mubadala will use it. But once big allocators put models on deal work, your data room has two readers. The partner reads the story. The model checks whether it adds up: deck revenue vs the monthly numbers, market size vs its source, cohorts vs the growth chart. A human skims past a gap between slide 4 and tab 3 of the model. A machine won't. My advice to founders raising in the Gulf next year: reconcile everything before you send the link. The clean companies will get faster answers.

Uber just showed what tokenized cash is actually for. Not trading. Weekends. In a pilot with HSBC, Franklin Templeton, Tradeweb and Fireblocks, it used tokenized bank deposits to buy a tokenized money market fund on a Thursday, then sold part of it on a Sunday. Normally that cash gets pulled before Friday's close. Per Bloomberg, this kept it earning about 1.5 extra days. Small for Uber. Not small here. A Saudi company paying in dollars works Sunday to Thursday. New York works Monday to Friday. That's four shared banking days a week, and treasuries hold idle buffers to cover the gap. This is the tokenization pitch I'd fund: a bank on one side, a regulated fund on the other, a CFO as the buyer. In the Gulf, the first real users won't be crypto funds. They'll be treasurers.

The number that matters in Broadcom's $63B Anthropic deal isn't 63. It's 72.5%. Banks and Blackstone fund an SPV that buys Broadcom chips and leases them to Anthropic. If the racks end up worth less than 72.5% of what's still owed on the lease, Broadcom covers the gap. Peak exposure: about $43B, running down to zero over five years of lease payments. The lease covers Anthropic's credit. The guarantee covers what nobody can price yet: what these chips are worth late in the lease. And it's not the first. On an earlier $35B Apollo/Blackstone deal, Broadcom's backstop was up to $29B. Moody's has already flagged the risk if these stack up. How I read it as an investor: • when the seller guarantees the floor, the seller is the real underwriter • in any compute or hardware pitch, I now ask who holds the residual, not just who signed the lease • when these SPVs get offered to Gulf allocators, the first question is which layer. The second lien sits under ~$43.5B of senior debt

An AI startup for wealth advisors just raised $70M at a reported $500M valuation. Two years old, about 20 customers. The number behind it isn't seats. It's assets on the platform: $50B to $700B this year, now adding ~$75B a month. Revenue up more than 25x (no figure given). That's the metric I'd underwrite in wealth tech. It grows when the customer's book grows, and once accounts open through you, you're hard to rip out. What I'd fund in the Gulf: • the boring middle: onboarding, KYC, suitability, custodian paperwork. Nevis says it cut account opening from weeks to days • built for DIFC and ADGM rules, Arabic and English, multiple custodians What I'd pass on: another meeting note-taker for advisors. Every CRM will ship that as a feature. DIFC alone has 592 wealth and asset managers and 1,408 family-related entities. Whoever builds the Nevis for them gets my meeting.

Two Series A rounds on Tuesday, both priced around $1.5B. • Arda (Bob McGrew, ex-OpenAI research chief): $160M at $1.5B. Already inside Anduril's manufacturing software • Eka Robotics (MIT + ex-DeepMind founders): $150M at $1.45B post. Robot hands that sense force Each sold roughly a tenth of the company. That's growth-round math with an A label. The price isn't the risk I'd underwrite. The next round is. A clean B usually wants a 2–3x step-up: $3–4.5B, earned on factory revenue, not résumés. From Dubai, this is the AI I'd rather own than another chat app. The Gulf is spending on factories and data centres. When an A is priced like a C, I ask one thing: what has to be true at the B? If the answer is "more of the same team," I pass.

Meta just printed a new currency: stock with no vote. Same quarter, free cash flow went negative for the first time in over a decade: • Capex $34.9B vs $32.8B of cash from operations • No buybacks this year • 2026 capex guided to $135–145B The part I'd copy into a term sheet is the staple. Zuckerberg can sell 20% of his new Class C freely. For the rest, each share sold generally has to be matched by selling a voting share or turning a 10-vote share into a 1-vote one. Gulf family groups heading to IPO want exactly this: outside money, inside control. When a founder asks me for super-voting shares now, I use the Meta test: fine, if cashing out costs you votes too.

Agree on the rails. The liquidity is the part nobody has built yet. Saturday night, with Wall Street shut, chip-stock perps fell 6–7% in ten minutes. ~$51.6M liquidated, ~$38M of it longs. With no cash price to anchor to, the contracts traded off each other. It hit at 01:22 Sunday, Dubai time. By Monday morning here, the Micron perp was back above where it started. When a tokenized-stock venue pitches me, I skip the chain question. I ask who makes the market at 1am on a Sunday, and with whose balance sheet. 24/7 trading is solved. 24/7 liquidity is the company I'd fund.

A round with a deadline is a round priced to dodge a comparison. OpenAI wants ~$30B at $1.4T+ committed before Anthropic's expected mid-November IPO, per Reuters. In March it was $852B. The catch is the yardstick: • OpenAI pitches $50B annualized revenue, booked net of partner shares • Anthropic was at ~$65B by July. OpenAI says it counts cloud resales gross; Anthropic says its accounting is standard Abu Dhabi's MGX is in talks on this round. It co-led Anthropic's Series G in February. Gulf money is the one buyer that doesn't have to pick a side. My rule when a founder sets a close date before the comp lands: I wait for the comp. If the price is right, it's still right in December.

The number I'd underline in Polsia's story isn't $25M. It's $1.2M. That was its monthly Anthropic bill earlier this year, right after projected revenue jumped to ~$10M a year. Annualized, that's ~$14M of tokens against ~$10M of revenue. Then Sapiom rerouted the traffic, mostly to open-weight models. The bill fell to ~$100K a month. Same product, one routing decision. The wider data agrees: a 10-point intelligence lead now comes with ~20% more developer traffic, down from ~69% in late 2025 (The Information). When an AI-native company pitches me, I ask for gross margin after inference before I ask for ARR. Then how fast they can switch models. A one-person company can get to $25M. A one-API company has a landlord.

$750B of Big Tech capex this year, up from ~$410B. Most people read that as demand. I read it as next year's comp. Same week, the Nasdaq-100 fell ~5.5% and the chip index ~11%. The market is already asking what +80% does for an encore. What it changes on my side of the table in Dubai: • "we sell picks and shovels into the build": I pass more often now. One budget review and the pipeline is gone. • "we make the GPUs they already bought earn more": inference cost, utilisation, AI that ships revenue for the end customer. That's where I lean in. The last cycle funded the build. The next one funds the payback.