The FourWeekMBA Daily — today’s AI moves, told through the Business Engineer lens.
Three deals on one day, three bets on where AI value concentrates once the model itself is rentable — each priced aggressively on durability that has not yet been proven.
Three deals landed on August 12 that, read together, form a single argument. Thrive Holdings — the vehicle Josh Kushner spun out of Thrive Capital — raised more than $2 billion at a self-reported $12 billion valuation to keep buying and operating fragmented, mission-critical services businesses. It already holds more than seventy firms in accounting and IT, and the strategy is explicit: deploy AI inside their workflows not to sell software, but to capture the productivity gain as the owner of the business. That inversion — AI-as-owner rather than AI-as-vendor — is the entire thesis.
On the same day, Ryanair signed a five-year agreement to move its 35,000 employees onto Google Cloud and Workspace, adopt Gemini Enterprise for custom agents, and run DeepMind’s AlphaEvolve and WeatherNext against crew logistics, maintenance scheduling, and weather disruption. The revealing detail is not the chatbot — it is AlphaEvolve and WeatherNext, narrow optimization and forecasting tools, that carry the operational ROI. Google won the deal by bundling infrastructure, productivity software, and frontier research into one commitment that is operationally hard to unwind.
And Stockholm’s Lovable raised $400 million at $13.3 billion — more than double its December 2025 mark — on approximately $500 million in annualized revenue run rate, with Workday, Asana, and Nvidia among its customers. Natural-language software creation turned out to be one of generative AI’s few genuine commercial hits. The caveat is structural: that $500M figure annualizes a single recent month, the multiple lands near 26×, and Lovable builds on foundation models it does not own — supplied by Anthropic and OpenAI, both of which are moving into coding tools themselves.
The key insight: All three deals are bets on the same underlying shift — as the model commoditizes, durable value moves off the model and onto the layers around it. But each deal prices that thesis at a mark that assumes the lead holds. The direction is proven. The durability is the wager.
The Map of AI framework organizes this cleanly. Thrive is playing at the deployment layer — owning the business unit that realizes the gain, rather than selling a tool to it. Ryanair-Google is the distribution layer — surfaces and integration depth as the moat, which is also exactly what the analysis of post-Nvidia value concentration predicted: once compute margins compress, the next defensible layer is controlled distribution. Lovable is the application layer — the place where, right now, AI revenue is actually showing up first. The problem is that “first” and “durable” are not the same claim.
Map of AI — Deployment Layer Logic
Once the model is a commodity, the profit pool moves to whoever controls the workflow, the surface, or the switching cost.
Thrive captures it by owning the firm. Google captures it by owning the integration. Lovable captures it by owning the interface — for now. The shared vulnerability: none of these positions is structurally sealed. Thrive has to execute AI deployment into messy legacy systems. Google’s bundle only holds while the specialized tools (AlphaEvolve, WeatherNext) outperform alternatives. Lovable’s interface advantage lasts only as long as its model suppliers stay upstream.
The honest separation is between two claims embedded in every valuation today. The proven part: AI value is visibly concentrating downstream of the model, in deployment, distribution, and applications. That is visible in the revenue, the customer lists, and the deal structures. The assumed part: that these specific positions hold their leads long enough to justify marks priced for a winner-take-most outcome. Thrive’s $12 billion is a private self-reported mark, not a cleared market price. Lovable’s 26× multiple rests on a single month annualized. Ryanair’s ROI from AlphaEvolve and WeatherNext has not been delivered yet — the contract is signed, the efficiency gains are not.
Three Implications
THE ROLL-UP PLAYBOOK GETS AN AI THESIS
Thrive Holdings is the clearest articulation yet of AI-as-owner: buy the business, internalize the margin expansion, skip the thin SaaS subscription. If the deployment execution works, this is a structurally superior capture mechanism to selling AI tools. But deploying AI into fragmented legacy accounting and IT workflows is slow, messy, and firm-specific — the operational risk is real, and the $12 billion self-reported mark will not be stress-tested until a liquidity event arrives.
ENTERPRISE AI ROI LIVES IN NARROW TOOLS, NOT GENERAL CHAT
The Ryanair deal makes this explicit. AlphaEvolve and WeatherNext — optimization and forecasting models with a specific operational mandate — are where the business case lives. Gemini Enterprise provides the productivity surface. The bundle wins the deal; the specialized tool delivers the ROI. That distinction matters for any enterprise buyer evaluating AI spend: the general assistant is the wrapper, the narrow operational model is the moat.
THE APPLICATION LAYER MONETIZES FIRST AND GETS SQUEEZED NEXT
Lovable is real revenue and a real product — but the wrapper-versus-platform tension is not theoretical. Anthropic and OpenAI are both moving into coding tools, and Lovable builds on their models. The application layer is where AI revenue is showing up first in 2026, which is precisely why the model providers want a share of it. A 26× run-rate multiple prices a durable lead; the competitive structure prices a race. Those are not the same scenario.
Business Engineer Framework
The Map of AI: Where Value Concentrates When the Model Is Rented
Thrive, Ryanair-Google, and Lovable each occupy a different layer of the AI stack — deployment, distribution, application. The Map of AI framework maps all nine layers across 200+ companies and shows exactly where margin pools are forming and where they are being competed away. Today’s three deals are case studies in real time.
August 12 was the day the AI value-capture argument got priced three ways simultaneously: own the business, own the distribution, be the breakout app. All three positions are structurally coherent — downstream of the commoditizing model is exactly where the profit pool is moving, and yesterday’s Gemini-distribution and Databricks-agent-layer stories said the same thing from the infrastructure side. What the market has not yet resolved is which of these specific bets holds its lead long enough to justify a self-reported private mark, a 26× run-rate multiple, or a five-year operational commitment with ROI still undelivered. The direction is the consensus. The durability is where the actual investment thesis lives — and none of today’s deals has proven it yet.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
