What the White House tech dinner means for finance
Here’s the AI stack that matters and what to actually fund:
Entry-level ladders are getting rebuilt. The only durable edge left for finance is judgment, data moats, and the speed to operationalize both.
And yes, policy and power matter.
We saw that at the White House.
The Night the House Aligned (Because of AI)
You’ve seen this photo.
A long table, high ceilings and careful smiles.
On one side: the architects of our digital life, platform founders, chip and cloud CEOs, and infrastructure builders. On the other: the seat of power. The headlines called it a tech dinner, but that undersells it.
I call it a coordination point.
Here’s what you don’t see in the photo:
The Oval Office ritual. Before dinner, the procession runs like a Swiss watch. Names announced, photos taken, a pen and a coin placed in each hand, the choreography of state meets the improvisation of startup life. The moment shrinks even the boldest personality. You’re not a headline in this room. You’re a participant.
The room between rooms. The Roosevelt Room becomes a holding pattern for titans. A few founders shoulder-to-shoulder with the quiet operators who actually turn policy into action. Everyone knows everyone. Nobody can hide behind a PR wall.
The table map. Not a printed list taped to a wall, but a model of the table with escort-level detail. Nothing about the seating is random. Proximity is a signal.
What struck me was really the alignment.
Competitors who would never share a product roadmap were, for one night, rowing in the same direction.
Make. Invest. Build. In America. Now.
The incentives, permits, energy policy, and export posture are (finally) bending in favor of heavy CAPEX and faster deployment. You can feel the center of gravity moving from talking about AI to installing AI.
This is what matters for CFOs: coordination points change time horizons.
When policy, capital, and capability snap into alignment, projects that were someday become this quarter.
Excel is Ending
For 40 years, spreadsheets were a ritual. You learned finance by suffering through it. Build the model. Break the model. Build it again. You earned judgment cell by painful cell.
AI just cancelled the suffering.
DCF in minutes. Data ingest in seconds. Reconciliations, joins, cleans, and lookups are offloaded to an agent that never tires and rarely complains. Google and Microsoft are embedding copilots. Startups are racing to wrap entire finance workflows in autonomous routines.
4 hours of work can be done in 20 minutes.
If you manage a finance team, then it’s a budgeting question. A hiring question. A power question. The org chart is about to change shape.
The market will not reward you for work AI can do.
The value shifts to what AI can’t do well.
Choosing assumptions
Constraining the model
Deciding the range
Telling the story
Changing the plan
Your new powers are judgment under uncertainty.
The ladder is shorter, steeper, and faster.
Your job as CFO is to rebuild it on purpose.
What Invest in AI Actually Means
Here’s the stack that matters and what to actually fund:
1) Compute & Power (Plumbing Layer)
What to fund: committed GPU/TPU capacity; multi-cloud contracts with burst rights; colocated racks near cheap, reliable power; power purchase agreements.
Why it matters: cost per inference and latency are strategic constraints. If you can’t get cycles, you can’t scale use cases. If you can’t predict your energy costs, your AI unit economics will wobble.
CFO move: treat compute like you treat debt. Ladder expirations, diversify providers, and negotiate floors/ceilings. Lock in allocations, not just discounts.
2) Data Quality & Rights (Moat Layer)
What to fund: data contracts with product and GTM; event schemas; lineage; PII vaulting; internal data marketplace; IP counsel for training rights.
Why it matters: the same model on different data ≠ the same output. Your moat is not the model; it’s permissioned context.
CFO move: create a data capitalization plan for what to build, what to buy, and what to license. Tie model performance to the cost of data acquisition and cleansing. Make data debt a line item.
3) Agentic Workflows (Throughput Layer)
What to fund: autonomous agents stitched to your systems (ERP, CRM, billing, data warehouse); guardrails; observation; rollback; human-in-the-loop gates.
Why it matters: chat is cute; agents move the needle. Close-the-books agents, AR collection agents, headcount-plan agents, and variance-explain agents. The wins stack when the agent writes to systems, not slideware.
CFO move: fund 3 factory lines of automation tied to P&L levers (Revenue, COGS, Opex). Each line ships quarterly and reports net efficiency per task (minutes saved × wage rate × error delta).
4) Apps & UX (Adoption Layer)
What to fund: durable interfaces where business users live, Sheets/Excel plugins, Slack bots, Notion wikis, and finance portals that auto-generate narrative.
Why it matters: the best AI is the one your team actually uses. Tools that piggyback on daily muscle memory have zero adoption tax.
CFO move: mandate that every AI initiative ships a self-serve UI and a change-management plan (training, playbooks, and prompts library). No UI, no budget.
5) Governance & Risk (License-to-Operate Layer)
What to fund: red-team testing, output audits, prompt logging, model cards, incident response, vendor DPAs, and domain-specific guardrails.
Why it matters: auditors and regulators won’t care that the model said so. You need traceability and human accountability.
CFO move: charter a Model Risk Committee with Finance, Legal, IT, and Audit. Set thresholds where AI can recommend vs. approve. Track AI incidents like you track SOX exceptions.
What the White House Night Really Signals
The dinner made 2 truths unavoidable:
The U.S. is re-industrializing around AI.
Data centers, fabs, grid upgrades, fiber, and logistics. These are not tech headlines; they are bond-sized projects. If your finance strategy doesn’t anticipate cheaper, closer compute and upgraded power, you’re planning yesterday’s cost curve.Coordination compresses time.
When government and industry sing from the same page. Even temporarily, permits move, exports open, standards solidify, and capital unlocks. That compresses the timeline from pilot to platform. If you wait for perfect certainty, you’ll be buying capacity from your competitor.
Behind the scenes, the message wasn’t AI will change everything.
We’re past that.
The message was:
Install capacity. Clear obstacles. Educate talent. Scale responsibly.
The 3 AI Bets I’d Make Before Year-End
If I had to place chips today, here’s where they’d go:
1: Agentic Finance Ops
Not AI in a chat window but agents with write access to ERP/CRM that own discrete workflows, billing reconciliation, PO processing, cash application, and headcount scenarioing. You don’t need artificial general intelligence. You need autonomy with guardrails and immaculate logging.
2: Data-for-Rights
Secure training rights for the proprietary data that makes your use cases sing, support transcripts, field notes, niche telemetry, and third-party long-tail datasets. Then build your context graph so models see your world the way you do.
3: Power-Adjacent Compute
Get closer to cheap power and predictable cooling. If you can’t own, reserve. Consider colocation near reliable generation. You will not regret over-planning capacity and latency for workloads that become mission-critical faster than procurement cycles can handle.
Supply is tight; demand is compounding.
Capacity is strategy.
The Bottom Line
AI will automate the median outcome.
That’s the point.
Let the machines own average.
The average model, the average narrative, the average plan. Great!
Your edge as CFO is everything above the median:
The assumptions you choose.
The constraints you enforce.
The context only your data reveals.
The story that aligns the org around what happens next.
The White House dinner was about picking up speed.
Government, industry, and capital are aligned to install the future, not debate it.
Move now, measure hard, and guard smart.
The spreadsheet era made great number-keepers.
The agent era will make great decision-makers.
But the real breakthroughs don’t happen in reports or Zoom calls.
They happen in real life, when 100 CFOs gather under one roof to share what’s working, what’s breaking, and what’s next.
On September 26th in Dubai, you can be in those conversations.
With only 100 seats, this event is filling fast.
Don’t wait for the memo.
Be in the room: https://luma.com/5udk521n
You can meet peers who are building AI-powered finance teams, not just talking about them. And you’ll walk away with strategies you can put into practice the next morning.
You have plenty of time to book your flights and lock in your spot.
And that’s all for today.
See you on Thursday!
P.S. I’ll be in Los Angeles this week for the All-In Summit. DM me if you're in town.
Whenever you're ready, there are 2 ways I can help you:
Promote your business to 24,000+ CFOs and senior finance leaders by sponsoring this newsletter.
If you’re building an AI-powered CFO tech startup, I’d love to hear more and explore if it’s a fit for our investment portfolio.
Advertise with the AI CFO Office.
I’m Wouter Born. A CFOTech investor, advisor, and founder of finstory.ai
Find me on LinkedIn