ExpertsFMMartin Casado on Frontier Model Economics, AI's Slowdown, and the Cursor Deal - [Invest Like the Best, EP.498]
Invest Like the Best with Patrick O'Shaughnessy · November 4, 2026
5 Selected Lessons:
framework · 3 heuristic · 2
Four investor mindsets
Source
Invest Like the Best with Patrick O'Shaughnessy — Martin Casado on Frontier Model Economics, AI's Slowdown, and the Cursor Deal - [Invest Like the Best, EP.498] (November 4, 2026) · at 0:00 ▸
When this applies
when deciding how to evaluate any investment or resource-allocation decision under uncertainty
The takeaway
Before judging any opportunity, explicitly name which lens you're using to assign value, because different lenses demand different evidence and conflating them produces incoherent decisions.
Why it works
Each mindset implies a different proxy for value (market/financials, intrinsic productive value, founder conviction, or price momentum), and mixing proxies lets people justify any decision post-hoc.
One level up
Self-awareness about your own decision framework prevents motivated reasoning dressed up as analysis.
Where it stops holding
This stops working when a decision genuinely requires blending lenses deliberately (e.g., late-stage deals where both financial health and disruptive potential matter) rather than picking one.
Detail
procedure:
- Identify if you are a financial investor (proxy = market size + SaaS metrics)
- Identify if you are a productive investor (proxy = technical approach, team, value-creation trajectory)
- Identify if you are a frontier investor (proxy = founder conviction, YOLO on team quality alone)
- Identify if you are a speculative investor (proxy = market timing, price momentum)
- Pick one lens and apply its diligence consistently; don't borrow criteria from another lens
benchmarks: {}
rationale: If you don't know what your beliefs are about value, you'll always justify whatever personal bias you have.
In their words
So, if you don't know what your beliefs are, you'll always justify whatever personal bias you have.
Moat mechanism vs. moat shape
Source
Invest Like the Best with Patrick O'Shaughnessy — Martin Casado on Frontier Model Economics, AI's Slowdown, and the Cursor Deal - [Invest Like the Best, EP.498] (November 4, 2026) · at 6:42 ▸ · at 10:08 ▸
When this applies
when assessing whether any claimed competitive advantage is actually durable
The takeaway
Separate 'what creates this advantage' from 'how does this advantage grow with scale' — defensibility claims are only meaningful once you've established the growth shape.
Why it works
People conflate the cause of an advantage (brand, scale, two-sided marketplace, network effect) with how that advantage scales mathematically (super-, linear, or sub-linear); only the scaling shape determines whether a leader pulls away or a challenger catches up.
One level up
Precise causal decomposition of vague strategic terms prevents false confidence in defensibility arguments.
Where it stops holding
This stops working when competitive equilibria shift the inputs (e.g., a better-funded rival), since the shape of one's own moat can be altered by competitors' actions.
Detail
procedure:
- Ask what mechanism produces the claimed advantage (network effect, scale effect, two-sided marketplace, brand)
- Separately ask whether that mechanism's value grows super-linearly, linearly, or sub-linearly with scale
- Only treat the moat as defensible if growth is super-linear
- Remember this is a competitive equilibrium problem, not an isolated calculation
benchmarks: {"superlinear_examples":"telcos, LinkedIn, GitHub","frequency":"Casado says true superlinear (Metcalfe-style) network effects are rare"}
rationale: You have this very common word that's on 70% of pitch decks ... and nobody has consensus on what it means.
In their words
I think the root of the confusion is conceptually we conflate the mechanism with the shape of the moat.
Strategic M&A requires base case + inductive step + trust, not auction tactics
Source
Invest Like the Best with Patrick O'Shaughnessy — Martin Casado on Frontier Model Economics, AI's Slowdown, and the Cursor Deal - [Invest Like the Best, EP.498] (November 4, 2026) · at 51:45 ▸ · at 55:27 ▸
When this applies
when trying to close a high-stakes deal with a highly sophisticated counterparty whose decision rests on believing in a future option, not just current financials
The takeaway
In high-stakes strategic deals with sophisticated counterparties, win through demonstrated trust and a credible growth narrative (an inductive proof of future value), not through competitive-pressure tactics, because the latter signal unreliability to parties who can see through games.
Why it works
A strategic buyer is purchasing an option on a large future market, not just current revenue, so the seller must construct a logical proof (base case of current traction + inductive step showing how it attaches to the buyer's platform to unlock a much larger market) — and because buyers are highly sophisticated, attempts to manufacture leverage via leaks/auctions/competitive bidding destroy the trust needed for them to believe that proof.
One level up
The right negotiating posture depends on whether you're trading a commodity (where leverage tactics work) or selling a thesis about the future (where credibility and trust are the real currency).
Where it stops holding
This stops working for more commoditized, auction-style M&A where buyers are evaluating financial metrics/spreadsheets rather than a strategic option on a market — Casado notes that type of deal has 'a very different flavor' and leverage tactics may be appropriate there.
Detail
procedure:
- Establish deep personal alignment with the buyer's leadership on vision/strategy before formal deal talks
- Present the base case: current traction and existing budget capture
- Present the inductive step: how attaching to the buyer's platform/distribution multiplies the addressable market
- Engage in good faith, without leaking to press or manufacturing competing bids
- Let an in-progress funding raise help set objective price/value signals
benchmarks: {"example":"Nicira/VMware: a $1B option on a $40B market"}
rationale: What the buyer is doing is they're basically buying an option on a large market... you kind of have to solve the inductive proof for them.
In their words
So what the buyer is doing is they're basically buying an option on a large market. So in the case of VM, we're buying Nasira, they're buying a $1 billion option on a $40 billion market.
Defensibility matters less in growing markets
Source
Invest Like the Best with Patrick O'Shaughnessy — Martin Casado on Frontier Model Economics, AI's Slowdown, and the Cursor Deal - [Invest Like the Best, EP.498] (November 4, 2026) · at 12:33 ▸
When this applies
when deciding how much to worry about competitive defensibility for any opportunity in an expanding market
The takeaway
Weight defensibility analysis by market growth stage: it's a second/third-order concern in expanding markets and a largely moot point in consolidating ones.
Why it works
In a growing market, companies expand into new white space rather than fight head-on, so competitive pressure that would erode an undefended position doesn't materialize; in shrinking/flat markets consolidation happens regardless of moat quality.
One level up
Match the sophistication of your risk analysis to the actual stage of the game you're in, not to generic finance doctrine.
Where it stops holding
This stops working once a market saturates or shrinks — then consolidation happens and lack of defensibility becomes decisive, or at later/liquid stages where you must consider quality of revenue (e.g., buying a late-stage company's shares).
Detail
rule: In a growing market don't over-weight moat/defensibility analysis; in a shrinking one, moat quality mostly won't save you either since consolidation happens anyway.
threshold: Market still has significant penetration headroom (Casado cites AI being far under 1% of enterprise budget) as the signal growth still dominates.
rationale: I remember early discussions on I can't tell you how many investors are like, I will never invest in a model. It'll never accrue value. It's not defensible. That's fine, but as long as the market's growing in a large market, it's going to accrue a lot of value.
In their words
If a market you're investing in is growing, then companies are not going to want to compete. There's like a lot of new white space that gets created.
Capital can now substitute for engineering bottlenecks in AI-era software
Source
Invest Like the Best with Patrick O'Shaughnessy — Martin Casado on Frontier Model Economics, AI's Slowdown, and the Cursor Deal - [Invest Like the Best, EP.498] (November 4, 2026) · at 37:19 ▸
When this applies
when deciding whether throwing more money at a problem will actually accelerate progress, versus hit organizational/engineering limits
The takeaway
Reassess whether a domain's historical 'you can't buy your way out of this' constraint (coordination/engineering limits) still holds once a new tool changes the production function — if it's been relaxed, capital becomes a legitimate lever again.
Why it works
Historically engineering speed scaled sublinearly with people added (mythical man-month, coordination overhead), making large capital injections unproductive; AI-driven tooling removes much of that coordination bottleneck, so capital can be converted into output more directly than before.
One level up
Major technology shifts can convert previously structural bottlenecks (organizational, engineering) into resource-allocation problems, fundamentally changing what 'scaling' means.
Where it stops holding
This stops working for problems still gated by physical-world constraints (hardware, power, materials, scientific unknowns) where Casado says it's unclear capital scales the same way.
Detail
rule: In the AI era, unlike pre-AI software, capital can be productively deployed by a small team to drive rapid growth because the engineering bottleneck (mythical man-month) is relaxed.
threshold: Casado cites giving 20 people a billion dollars as now plausibly productive versus 'they would detonate' a decade ago.
rationale: In the history of tech, you haven't been able to do that because it was always an engineering problem, which just doesn't scale.
In their words
I really think the big disruption of AI is the fact that you can throw capital at problems. In the history of tech, you haven't been able to do that because it was always an engineering problem, which just doesn't scale.