The author walks through a16z's portfolio company by company — defense contractors like Anduril and Shield AI, surveillance firms selling face recognition and predictive policing, crypto rails shifting toward political influence, and labor-replacing AI agents. The core argument is that while any single bet might be defensible, the aggregate composition reveals a coherent thesis: a future that is militarized, monitored, financialized, and staffed by fewer humans.
The editorial argues that because venture capital sets the technology agenda far beyond its size, when a top-tier firm concentrates capital in a theme, founders pattern-match to it, engineers get hired to build it, and the discourse treats it as inevitable. This makes a16z's portfolio composition a leading indicator of what future job descriptions and consumer products will contain, not merely investment gossip.
Frames a16z's evolution as a betrayal of its own decade-long branding as the intellectual home of consumer and enterprise software. Points to the shift toward autonomous weapons startups, municipal surveillance contracts, and crypto positions that have quietly pivoted from consumer apps to political influence machinery as evidence the firm's public narrative no longer matches where the checks are actually going.
A long-form piece on modelrepublic.org went to the top of Hacker News with a blunt thesis: Andreessen Horowitz, the venture firm that spent a decade branding itself as the intellectual home of software, is now writing its biggest checks into defense contractors, surveillance infrastructure, crypto rails, and AI systems whose explicit selling point is doing the work of people who currently have jobs.
The author walks the portfolio company by company. Anduril, Shield AI, and a cluster of adjacent autonomous-weapons startups get flagged first. Then the surveillance layer — companies selling face recognition, license-plate readers, and predictive policing to municipalities and federal agencies. Then the crypto positions, which the piece argues have quietly shifted from consumer apps to political influence machinery. Finally the labor-facing AI plays: agent startups pitched to CFOs on headcount reduction, voice-AI companies replacing call-center workers, and vertical SaaS whose deck explicitly promises to eliminate a category of role.
The argument isn't that any single bet is indefensible — it's that when you zoom out on the whole fund, the picture is coherent, and the coherence is bleak. The piece frames a16z less as a diversified tech investor and more as a directional bet on a specific kind of future: militarized, monitored, financialized, and staffed by fewer humans.
Venture capital is a small industry that punches far above its weight in setting the technology agenda. When one of the two or three most influential firms concentrates capital in a theme, three things happen in sequence: founders pattern-match and pitch that theme, engineers get hired to build it, and the discourse — conferences, podcasts, Twitter — starts treating it as inevitable. This is why the portfolio composition question is not gossip. It's the leading indicator for what your next job description will contain.
The piece is strongest when it stops moralizing and just tallies. The defense allocation alone is a meaningful percentage of recent fund deployment, which is a genuine departure from the firm's 2010s posture. Whether that's a good thing depends on your politics, but the shift itself is not really contested. Marc Andreessen has been publicly explicit about it in his 'Techno-Optimist Manifesto' and subsequent essays, and the firm launched a dedicated American Dynamism practice specifically to fund defense and gov-tech. The critique in the modelrepublic piece is that the manifesto's soaring language ('we believe in adventure', 'we believe in ambition') sits awkwardly next to a portfolio whose most concrete products are targeting systems and workforce-elimination tools.
Smart people can disagree about whether Western democracies need a re-industrialized defense base — many serious engineers now believe they do — but you can't have that argument honestly while pretending the products are something other than what they are. The value of the modelrepublic audit, even if you reject its conclusions, is that it forces the specificity. Anduril builds autonomous weapons. Palantir-adjacent portfolio companies build surveillance. Sierra and its cohort exist to reduce headcount in customer-facing roles. Say the sentence out loud, then decide how you feel.
The crypto section of the critique is the weakest, and it's worth noting. The piece leans on the political-influence angle — Fairshake, the crypto super PAC, and a16z's public lobbying posture — more than on the technology, and reasonable people can argue that political engagement is a legitimate response to hostile regulation rather than evidence of bad faith. If you're skeptical of the whole critique, this is where the skepticism has the most to work with.
Most of what a16z funds, you will encounter as a developer whether you want to or not. A partial list of a16z-backed tools that have shown up in mainstream engineering workflows in the last eighteen months: several of the top AI coding assistants and agent frameworks, a handful of the observability and infra companies you're evaluating, a majority of the 'AI voice agent' pitches landing in your inbox, and a nontrivial share of the vector-database and RAG-tooling layer. The portfolio isn't a fringe corner of the industry — it's a large fraction of the tools your CTO is being pitched this quarter.
The practical move isn't to boycott anything — it's to notice when a tool's roadmap is being driven by the incentives of its cap table rather than the needs of its users. A funded-to-eliminate-headcount company will ship features that make the elimination story more credible to buyers, even when those features degrade the experience for the practitioners who actually use the product. A defense-adjacent infra company will accept contract terms and data-handling requirements that leak into how the product is architected for everyone else. This isn't conspiracy — it's just how startups work when the biggest checks come with the loudest opinions.
For engineers making career choices, the map is genuinely useful. If you want to work on defense tech, the funding is there and the tempo is real; American Dynamism is not a marketing bit. If you don't, know which logos to filter out of your job search, because the branding often obscures the underlying product. And if you're a founder pitching a16z in 2026, understand that the fund's thesis has narrowed. 'Neutral developer tools' is a harder sell than it was in 2019.
The modelrepublic piece will get dismissed in some quarters as ideologically motivated, and parts of it are. But the specific factual claim — that a16z has, over the last three years, concentrated capital in defense, surveillance, crypto-political-infrastructure, and labor-replacing AI — is basically correct, and the firm itself would mostly agree with the taxonomy while contesting the adjectives. The real question isn't whether the portfolio is bleak; it's whether the rest of the industry is going to keep letting one fund's thesis define what 'the future of software' means by default. Engineers, of all people, have the leverage to answer that with their labor. Worth knowing where the money actually goes before you decide.
> Lee's stated goal was to “desensitize everyone to the phrase ‘cheating.’” If you say it enough, he argues, “cheat begins to lose its meaning.” A16z praised Lee's approach as “rooted in deliberate strategy and intentionality.”Investors: but surely they won't cheat me!
I know this is a very minor point in the article, but reading the message next to the Doublespeed phone farm:> out of 130 dms sent, 15 pointed to a conversion.> From a former ecom guy's lens this is insaneWhy do these otherwise very smart people always seem to draw these kind of shortsigh
How come Flock is not listed? That’s the most troubling company a16z has heavily invested on and supported.
Nice hit piece.ModelRepublic.org is run by Tyler Johnson and funded by Dustin Moskovitz.A16z funds small tech. This includes companies that do Crypto and AI. You may not like it, but these are products and services that consumers want to use.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
"A man who wishes to make a profession of goodness in everything must necessarily come to grief among so many who are not good. Therefore, it is necessary for a prince, who wishes to maintain himself, to learn how not to be good, and to use this knowledge and not use it, according to the necess