Anthropic just shared which jobs AI will replace first. CFOs, finance tied for number one. Every CFO must see this chart
Jensen Huang says employees need to spend half of their salaries on AI tokens. Sequoia says the next trillion-dollar company won’t sell software.
AI is now everyone’s problem. But you can catch up.
I spent this week working with my small CFO research team, focusing on three things that changed how I view the future of finance.
Anthropic published the most detailed study ever done on which jobs AI will replace first.
Sequoia Capital published a thesis that reframes how every company will make money.
Jensen Huang went on the All-In podcast and said something that should be on every CFO’s whiteboard by Monday.
In today’s letter I stitch it all together and tell you what it means.
If you’re a CFO, Controller or FP&A manager … this is highly relevant to you.
Let’s dive in.
Anthropic just shared which jobs AI will replace first
Every CFO must see this chart. Blue is what AI can do to your function. Red is what it's doing right now. The gap in finance is massive.
94.3% of what CFOs and their teams do every day can be automated. The highest of any profession they studied. But only 28% is being automated today.
That means the wave hasn't hit yet. It's coming.
Among 22- to 25 year olds in exposed roles, hiring already dropped 14%. The junior analysts, the entry-level accountants, and the first-year FP&A hires. They're disappearing and nobody is replacing them.
Then Anthropic did something interesting.
The same week they published the warning, they invested $100 million to train Deloitte, Accenture, Cognizant, and Infosys on Claude.
30,000 consultants at Accenture alone.
The company building the tool is telling you it will reshape finance. Then they're teaching every consulting firm how to bring it to your door.
Their researchers called it a possible,
Great Recession for white-collar workers.
The full study is worth reading.
So what matters for CFOs is the 94.3% number, which is the ceiling.
The 28.4% number is where we are right now.
Everything that happens next happens in that gap.
Sequoia says the Next Trillion-Dollar Company won’t sell Software
For every $1 spent on software, $6 is spent on services
Services: The New Software makes an argument that changes how you think about every vendor, every hire, and every budget line in your finance function.
A company pays $10K a year for QuickBooks and $120K for an accountant to close the books. The software was never the expensive part. The work was. Sequoia’s bet is that the next legendary company won’t sell software. It will just do the work. They call this trend the shift from “copilots” (tools that help professionals work faster) to “autopilots” (systems that deliver the completed outcome).
Accounting and audit alone is a $50 to $80 billion market.
Management consulting is $300 to $400 billion.
These aren’t software markets. They’re services markets that software could never touch because the work required human judgment.
AI changes that math.
And the structural tailwind is real. The US has lost over 340,000 accountants in five years. 75% of CPAs are nearing retirement. CPA exam candidates dropped 30%. AI isn’t just competing with labor.
It’s filling a vacuum that labor can no longer fill.
Employees need to spend half of their salaries on AI tokens
Jensen Huang went on the All-In Podcast on March 19, taped live at NVIDIA’s GTC conference. He said something that should reshape how every CFO thinks about headcount budgets.
Let’s say you have a software engineer or AI researcher and you pay them $500,000 a year. That $500,000 engineer, at the end of the year, I’m going to ask them how much did you spend in tokens? If that person said $5,000, I will go ape. If that $500,000 engineer did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed.
He compared it to a chip designer refusing to use CAD tools. Doing your job without AI is like saying,
“I’m going to use paper and pencil.”
Think about what that means for finance.
If you have a $200K senior FP&A analyst and they’re spending $500 a year on AI tools, you are massively underinvesting in their productivity.
Huang’s framework says AI token spend per employee should approach 30 to 50% of their compensation. Most companies are in single digits.
Huang’s broader thesis is that computing demand has increased 10,000x in just two years. 100x from generative to reasoning AI, and another 100x from reasoning to agentic AI. NVIDIA’s fiscal year 2026 delivered $215.9 billion in revenue, up 65% year over year, with a $500 billion order backlog.
His most counterintuitive claim challenges the
AI kills enterprise software narrative.
He argues the opposite.
Every enterprise software company is about to become a distribution channel for AI tokens. Instead of being limited by the number of humans using the software, you’ll have 100 agents per employee banging on those tools.
Three other things Huang said that I keep thinking about:
Things that, ‘wow, this is too hard,’ that thought is gone. ‘This is going to take a long time,’ that thought is gone. ‘We’re going to need a lot of people,’ that thought is gone. You’re reduced to creativity. I think every engineer is going to have 100 agents.
And on whether AI spending is a bubble: asked about Dario Amodei’s forecast of $1 trillion in AI revenue by 2030, Huang said he’s “very conservative” because Amodei hasn’t accounted for enterprise software companies becoming token resellers.
