Home

 › 

Uncategorized

 › 

a16z’s David George Says $40 Billion Makes an AI Model Much Better. He Says the Same Money Would Have Done Nothing for Salesforce.

Ink illustration: two wooden pedestals with identical piles of coins at their bases. On the left, a pipe feeds coins into a glass sphere glowing orange; on the right, a plain gray tray sits unchanged.

a16z’s David George Says $40 Billion Makes an AI Model Much Better. He Says the Same Money Would Have Done Nothing for Salesforce.

Quick Read

  • A top a16z investor says $40 billion would transform an AI model but destroy a company like Salesforce, and his reasoning exposes a fault line running through every big tech investment right now.
  • George defended the Vision Fund's core concept as sound, yet the same logic he uses to rescue it is the exact logic that could unravel his AI thesis.
  • The company George chose to prove AI is different is quietly dismantling his own example.
  • Building a portfolio and living off one are two completely different skills, and almost nobody teaches the second. This problem is what The Definitive Guide to Retirement Income helps, and it is free today. Read more here. (Sponsor)

Where we heard itUncapped with Jack Altman, re-released on The a16z Show as “David George & Jack Altman on AI, Autonomy, and the Next $25 Trillion.” Released Oct. 4, 2026. Speaker: Andreessen Horowitz general partner David George.

Andreessen Horowitz investor David George says that the AI cycle breaks a rule from every earlier technology wave. On Jack Altman’s Uncapped podcast, which his firm re-released on The a16z Show, George said that $40 billion toward training AI models makes them much better, and he called that relationship unique to now.

Quote: “If you put a $40 billion fundraise into Salesforce, what are they going to do? Go hire a bunch of sales reps and engineers? It just wouldn’t work.” David George, Andreessen Horowitz, on Uncapped with Jack Altman
My Investing News

Why Capital That Turns Into Product Changes the Math

In most technology cycles, money buys growth indirectly. A company raises a round and spends it on sales headcount, marketing, and expansion, but on the day funding arrives, the product is unchanged. George says AI models work differently. Model quality scales with training investment, so capital turns more directly into a better product. For readers outside venture capital, the takeaway is that the size of a check now shapes how good the product is.

He then picked a foil. George asked what SalesforceCRM would do with a $40 billion capital raise. He suggested it would hire sales reps and engineers, and said that would not work.

Sizing George’s Hypothetical Against Salesforce

The segment never explained that figure. Salesforce runs the largest customer relationship management platform and has a market capitalization of $193.15 billion. The raise George describes is $40 billion, a substantial portion of the company’s market value.

Bar chart: $40 billion is about a year of Salesforce sales. Salesforce’s market value was $193 billion at the Oct. 2, 2026 close, its revenue was $41.5 billion in the fiscal year ended January 2026, George’s example raise is $40 billion, and Salesforce’s share buyback under way in 2026 is $25 billion.
My Investing News

Salesforce Is Already Repositioning Around AI

The company George chose as his pre-AI foil is itself rebuilding around AI. According to its SEC filing, Salesforce “helps organizations of any size become agentic enterprises – integrating humans, agents, apps, and data on a trusted, unified platform.” The company has also organized its products around Agentforce Apps.

George’s core claim survives this. His point is whether extra capital improves the product. A software company adding AI features runs on different economics than a lab whose model quality scales with compute spending. Salesforce’s shift shows the line between AI companies and software companies is blurrier than his contrast suggests.

Salesforce’s second quarter, reported Aug. 26, 2026

Revenue$11.3BUp 11% from a year earlier
AI and data recurring revenue$3.9BAgentforce and Data 360, up over 210%
Agentforce Apps$7.2BTwo-thirds of subscription revenue
Contracted revenue$33.5BDue in the next 12 months, up 14%

A Returns Claim From a Growth-Stage Investor

George also cited internal Andreessen Horowitz analysis finding that half of private market returns come from Series C and later rounds. As companies stay private longer, that split could move to 70-30 in favor of late-stage investing, he warned. This is his firm’s internal work, and Andreessen Horowitz invests at the growth stage, so it has an interest in that conclusion.

His Vision Fund Defense Uses the Same Logic

Host Jack Altman posed a question. The question was whether SoftBank’s Vision Fund failed because it put too much money into companies or because it arrived too late in the cycle. George said both. He then maintained the fund’s concept as sound. It launched at the wrong point in the cycle, when more money did not make companies better.

That is his AI argument in reverse. George holds that capital improves the product now and did not then, so whether the Vision Fund was wrong or simply early depends entirely on whether that claim holds. His AI thesis and Vision Fund defense rest on the same premise. If the link between capital and product quality weakens, both look different.

Line chart: The SaaS names George cited trailed the market. $100 invested on Oct. 2, 2025 was worth $116 in the S&P 500 (up 16.2%), $99 in Salesforce (down 0.9%), $80 in Workday (down 19.8%) and $74 in ServiceNow (down 26.2%) on Oct. 2, 2026, dividends included.
My Investing News

Contrarianism Through Conviction

Altman cited a line he thought he may have first heard from Thrive. One way to be contrarian is to believe something nobody else believes. Another is to believe what everybody believes, but 10 times more strongly. George said it was close to his firm’s growth-stage thesis.

Believing the consensus more deeply than anyone else pays enormously when the consensus is right and leaves no hedge when it is wrong. That fairly describes concentrated late-stage AI investing right now.

The bottom line

George’s case is that AI rewards capital in a way software never did: more money for training makes the model itself much better, while the same $40 billion would do little for a company like Salesforce. His firm’s late-stage bets and his defense of the Vision Fund rest on that idea. The risk is that if extra spending stops making models better, both arguments weaken together.

To top