Steven L

finance. tech. wine. life.

[Peter Berezin @PeterBerezinBCA]

The AI bubble is primarily an earnings bubble rather than a valuation bubble. My report this week discusses the metrics investors should monitor to know when this bubble is about to burst.

Clients can read it here:

https://t.co/nPpZ5E1mas

[Marcelo P. Lima @MarceloLima]

The Benedict Evans interview on @stratechery laid out the thesis for Aggregation Theory in enterprise software:

The core tenet of Aggregation Theory is that the Aggregator wins by commoditizing the suppliers, forcing them to compete on price and efficiency.

The interview explicitly argues that AI models are becoming undifferentiated utilities, comparable to electricity or cloud hosting.

Evans argues that for enterprise buyers, the underlying model is irrelevant infrastructure.

"As an enterprise, I don't care if that runs on Anthropic, or Google, or OpenAI, any more than I care whether it runs on AWS, or GCP... that's the software company's problem."

Just as an Aggregator forces suppliers to become modular inputs (like a driver for Uber), Salesforce treats the AI model as a modular engine. This strips the model providers of pricing power and leverage.

Aggregation Theory states that value accrues to the entity that controls the direct relationship with the user. In the enterprise, this relationship is secured not just by software, but by liability, support, and trust.

Evans explains that enterprises need accountability. A law firm or Fortune 500 company won't buy a raw API key from OpenAI; they need a vendor to sue if things go wrong.

"No lawyer is going to go and buy an AWS API key... they're going to buy a product with a Salesforce and a support team and a throat to choke."

By sitting between the raw intelligence (the model) and the user (the enterprise), Salesforce creates a moat. They own the demand, allowing them to swap out the underlying supply (the model) at will.

The interview highlights a critical advantage Salesforce has over the models: Context.

The "Blind" Model: Evans notes that models are trained on public data and have "no idea what's inside Salesforce" or "what's inside McKinsey."

The "Stored Process": Salesforce owns the "institutional knowledge"—the actual workflows and proprietary data of the business. The AI model is just a generic reasoning engine; it is useless without the context that Salesforce controls.

[Steven L @stevenssoapbox]

@JerryCap Could you share which Ben Evan’s interview this references?

[Jerry Capital @JerryCap]

Ben Evans on Software

[Aakash Gupta @aakashgupta]

Sam just told every SaaS CEO their per-seat pricing model has an expiration date.

Read what Frontier actually does. It connects to your CRM, your data warehouse, your ticketing tools, your internal apps, then lets AI agents execute workflows across all of them. Salesforce charges per seat. Workday charges per seat. ServiceNow charges per seat. If an AI agent can run a sales workflow without a human ever logging into Salesforce, the entire economic logic of "per-seat licensing" collapses.

Salesforce knows this. They've been sprinting to build Agentforce and hit $1.4B in ARR across their agentic products, processing 3.2 trillion tokens, signing 18,500 deals. They're trying to be the agent platform before someone else makes them the agent's tool.

Now OpenAI is saying: we'll be the layer that sits above all of your enterprise software. We'll give agents identity, permissions, and access to every system. And we'll let third-party agents from Google, Anthropic, and anyone else run on it too.

That last part is the move. By going model-agnostic and vendor-agnostic on agents, OpenAI is positioning Frontier as the operating system of the enterprise. The SaaS vendors become the apps.

Fidji Simo said it herself. When she was CEO of Instacart, her team spent months integrating each AI tool, and they still ended up with silos. Every enterprise CTO has lived this exact pain.

The companies signing up tell the story. Intuit, Uber, State Farm, Thermo Fisher. These aren't AI experiments. These are Fortune 500 operations teams saying "we need one platform to manage all our agents."

OpenAI lost $5B+ last year. This is how they plan to make it back. Consumption-based enterprise contracts where agents run on your infrastructure, processing your data, 24/7. The revenue per customer isn't a seat license. It's metered compute on mission-critical workflows.

If Frontier works, every SaaS company becomes a feature inside OpenAI's platform. If it doesn't, OpenAI just handed Salesforce and Anthropic the playbook for what enterprises actually want.

Either way, "per-seat" is dying. The only question is who collects the compute bill.

[Rohan Paul @rohanpaul_ai]

"AI is a massive TAM (Total Addressable Market) accelerator for traditional SAAS companies."

It will do so by automating the 95% of "drudgery" work.

~ Jeff Horing, founding partner of Insight Partners, one of the world's largest venture capital firms.

[Nick Dorsey @Midnight_Captl]

Jensen tonight at the Cisco AI summit on how the market is getting it wrong with Software

Software = tools, AI will be maximally efficient by using the tools (tool use), not reinventing them

[SouthernValue @SouthernValue95]

Jensen on software: “Remember what software is. Software is a tool. There is a notion that software is in decline and will be replaced by AI. It is the most illogical thing in the world and time will prove itself... Suppose we are the ultimate AI, artificial general robotics – the ultimate version of AI that can solve any problem. If you are a human or robot would you use a screwdriver, or invent a new one? Id just use one. Would you use a chainsaw or invent a new chainsaw. You would use one. If you are artificial general robotics you will obviously use tools. Now do the digital version of that… if you are AGI would you use the digital tools like SAP and ServiceNow and Cadence and Synopsis, or would you reinvent them? Of course you’d use them. That’s why the latest breakthroughs in AI are… what… tool use... because the tools are designed to be explicit. There are many problems in our world where F = MA. You don’t need approximately, statistically, you need exact. So I think we want AGI to use tools.” $NVDA $SAP $NOW

[TBPN @tbpn]

Box CEO @levie's defense of software over vibe-coded, n-of-1 internal tools:

"If you're Ford, and you're doing your supply chain on an ERP system, you want that to work the exact same way every single time."

"The billions of transactions going through that ERP system, you cannot take for granted. So the idea that you're going to go vibe-code that is not possible, or at least not likely."

"The other point is: your company has a fixed amount of IT resources. And you have to decide what you're going to spend your time on as an organization."

"Do you want to spend time on rebuilding something that the market can supply you, that's seen best practices thousands of times? Or do you want to go and build that out with your n-of-1 experience?"

"Or do you want to spend your limited, scarce resources on building software, and building experiences, that will make you more money, and that will actually be used by your customers?"

"I think on the margin the average enterprise is going to spend their time and energy on the latter."

"I'm 100% bullish on vibe-coding, 100% bullish that we're going to have 100x more software. But that still doesn't cross the threshold where I would want to go build our own CRM system."

[market participant @undrvalue]

https://t.co/p8iqGRq1Td

[Aakash Gupta @aakashgupta]

Goldman just told every SaaS CEO their business model has a five-year shelf life and the market hasn’t repriced accordingly.

The headline number is $780 billion in application software by 2030, 13% CAGR. Sounds like growth. But agents capturing 60%+ of that economics means the profit pool migrates away from per-seat subscriptions toward workflow-completion pricing. The market gets bigger while the legacy revenue model gets smaller. Two things happening at once.

This is already showing up in the data. Seat-based pricing dropped from 21% to 15% of SaaS companies in just twelve months. Hybrid pricing surged from 27% to 41%. Klarna doubled revenue per employee after deploying agents across core workflows. SaaStr is actively downgrading seat counts at vendors because they have 12+ AI agents in production replacing human users.

The math problem for incumbents is brutal. Salesforce charges up to $500/seat/month at top tiers. When one agent automates what ten humans used to do, charging per seat becomes a penalty on the vendor. BCG’s buyer survey found 40% of enterprise customers cite seat reduction as their primary lever to cut software spending. The very AI features vendors are building to retain customers are giving those customers the tool to shrink their contracts.

ServiceNow saw this coming and pivoted to “AI Control Tower” positioning, generating $600 million from Now Assist in Q4 alone. But even with 21% subscription growth and 25% more monthly active users, the stock dropped double digits after earnings. The market is saying: prove the new pricing model scales before we assign a multiple.

Goldman’s own behavior tells the real story. They announced thousands of autonomous AI coding agents working alongside 12,000 human developers, projecting 3-4x productivity gains. Goldman is simultaneously publishing the research that says agents eat SaaS economics while deploying agents internally to eat SaaS economics. They’re the customer proving their own thesis.

The vendors who win will be the ones who wrap workflows in agents and price on outcomes, capturing a share of the productivity gain rather than passing it all through. The vendors who lose will be the ones still counting seats while their customers count agents.

[Aaron Levie @levie]

AI coding will lead to 100X more software. That to me doesn’t conclude that every customer wants to be responsible for owning and managing their own ERP. It concludes that customers will have more choice, better features, and far more personalization on top of software.

[Steven L @stevenssoapbox]

@NolanAnalyst Does it ever bother anyone how that lady just floats away? What kind of exit is that?

[Marc Benioff @Benioff]

LLMs are the new disk drives: commodity infrastructure you hot-swap for whoever’s cheapest + best. The fantasy that the model is a moat just expired. ☠️

[Palmer Luckey @PalmerLuckey]

“In keeping their options open, they ensure that they’re going to jump from option to option. If you don’t commit to a path, you’re going to fail at it … You have to commit to it to make it work, and I think marriage is the same way. You just have to commit to it. You have to say, ’This is the path I’m on. For better or for worse, I’m going to double down on it.’”

[Steven L @stevenssoapbox]

@MUFC_Timm He has MJ shooting leftie

[Steven L @stevenssoapbox]

@pitdesi Juniper Cafe croissants. They supply the croissants at StFrank’s in Soma and they’re incredible.

[Steven L @stevenssoapbox]

@TrungTPhan Is there not a section 382 equivalent for this?

[Antonio García Martínez (agm.eth) @antoniogm]

My Singapore driver informs me that robbery carries a minimum sentence of three years in jail and six strokes with a cane.

“One stroke almost kill you.”

“Other country you pass out drunk on the street…phone and wallet gone. In Singapore, nobody touch them.”

No shit.

[Steven L @stevenssoapbox]

@Norseman1 I remembered this from last week - this was a great call.