frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

LLMs feel more like CPUs than applications

1•derverstand•1h ago
I’ve been thinking about the current LLM wave and the historical microchip transition.

Microchips were never “products” in themselves. They were compute primitives. The real value emerged in operating systems, developer tools, and applications built on top.

Today, LLMs increasingly feel similar. OpenAI, Anthropic, Google, etc. are building cognitive compute layers. Most “AI startups” look like early PC software – wrappers around a new primitive.

Agents then feel like early operating systems: orchestration layers around probabilistic compute, adding memory, tool access, execution loops.

If this analogy holds, the long-term value might not sit in the base models themselves, but in: - workflow integration - vertical domain systems - data pipelines - distribution - orchestration layers

The open question: Is this closer to the Intel/AMD era (infrastructure shift), or something fundamentally different because the primitive is stochastic and language-native?

Curious how others see the analogy.

Comments

0xecro1•1h ago
It's fundamentally different, and that's exactly why the CPU analogy has an expiration date. Intel never built Windows. The chip layer and the application layer were structurally separated — different companies, different business models, different competencies. That separation is what created the massive ecosystem on top. Frontier labs aren't staying in the chip layer. Anthropic ships the model, the API, artifacts, computer use, MCP — primitive through orchestration through distribution. OpenAI has the model, ChatGPT, plugins, the app store play. They're vertically integrating the entire stack at a speed Intel never could, precisely because the primitive is probabilistic and language-native. When your compute layer already "understands" the task, the gap between infrastructure and application collapses. So short-term, yes, it looks like Intel/AMD. Base models commoditize, value flows to tooling and verticals. But long-term, the labs that own the primitive will likely own most of the stack above it too. The "just build on top" window might be shorter than people expect.
derverstand•35m ago
Fair point on the vertical integration. That’s definitely a real difference from the CPU era.

At the same time, I’m not fully convinced that owning the primitive automatically means owning most of the value on top of it in the long run.

Nvidia owns the GPU layer, but it doesn’t own the majority of the software built on top of GPUs. AWS owns infrastructure, but SaaS value still fragments heavily across vertical domains. Infrastructure providers often try to move up the stack, but specialization and domain depth tend to create space above them.

It’s possible that frontier labs will capture more horizontal value than chip companies ever did. But I’m not sure language-native compute completely collapses the stack. It might just reduce friction and lower the barrier for building vertical systems.

The interesting question to me is whether this really eliminates the ecosystem layer — or just reshapes it.

Git for Data Applied: Comparing Git-Like Tools That Separate Metadata from Data

https://motherduck.com/blog/git-for-data-part-2/
1•articsputnik•40s ago•0 comments

Show HN: Glass Box: writing editor that exports a verifiable PDF of your process

https://firl.nl/glass-box
1•normanbell•50s ago•0 comments

Section 230 Preempts Lawsuit over Unwanted Gmail Spam–Dor vs. Google

https://blog.ericgoldman.org/archives/2026/02/section-230-preempts-lawsuit-over-unwanted-gmail-sp...
1•hn_acker•1m ago•0 comments

I built a scanner to find common vulnerabilities in AI-generated apps

https://securemyvibes.com
1•baptisteallain•1m ago•1 comments

DOJ withheld Epstein files related to allegations Trump sexually abused a minor

https://www.npr.org/2026/02/24/nx-s1-5723968/epstein-files-trump-accusation-maxwell
3•pickleglitch•2m ago•0 comments

Bithumb $44B Bitcoin blunder puts South Korea regulators on alert

https://www.coindesk.com/policy/2026/02/09/usd44b-bitcoin-blunder-puts-south-korea-regulators-on-...
1•PaulHoule•2m ago•0 comments

ArcGIS Maps SDK for JavaScript – Version 5.0 Release

https://community.esri.com/t5/arcgis-javascript-maps-sdk-blog/arcgis-maps-sdk-for-javascript-vers...
1•Noash•2m ago•0 comments

Moody's alert cites gap in data centre accounting for Big Tech companies

https://www.ft.com/content/3ff9a481-8be3-4dd4-9a00-ea809f3485fd
1•petethomas•2m ago•0 comments

We Solved Execution. Now What?

https://anirudharamesh.github.io/blog/we-solved-execution-now-what/
1•anibot•3m ago•0 comments

Anthropic's new AI tool can write COBOL, sending IBM's stock tumbling

https://www.tomshardware.com/tech-industry/big-tech/ibm-stock-takes-a-13-percent-whiplash-after-a...
1•cdrnsf•3m ago•0 comments

Show HN: Stupid simple e-ink RSS reader

https://github.com/edleeman17/E-Ink-RSS-Reader
1•ed1727•3m ago•0 comments

Dev jobs up 10% YoY while other jobs down 5.8%. What do you see on the ground?

1•mrborgen•3m ago•0 comments

The DIY OpenClaw Assistant You'll Want to Carry

https://www.hackster.io/news/the-diy-openclaw-assistant-you-ll-actually-want-to-carry-97888b183ac1
1•toomuchtodo•3m ago•0 comments

The Secret History of Knocking on Wood: Most of human nature is not written down

https://resobscura.substack.com/p/neolithic-habits-machine-age-tools
1•benbreen•4m ago•0 comments

Show HN: Tiny-parquet – JavaScript lib to read/write Parquet in 326KB of WASM

https://github.com/nktrchk/tiny-parquet
2•Nikitaita•6m ago•0 comments

Is it just me or is reviewing PRs getting exponentially harder?

https://www.bitarch.dev/blog/the-hidden-cost-of-ai-assisted-coding
1•birdculture•6m ago•0 comments

LLM-as-a-Judge: Evaluating Output Without a Ground Truth

https://www.kerno.io/blog/llm-as-a-judge-evaluating-output-without-a-ground-truth
1•karimtr•6m ago•0 comments

Private Markets Hiring Defies Gloom with $2.5M Pay Deals

https://www.bloomberg.com/news/articles/2026-02-24/private-markets-hiring-defies-gloom-with-2-5-m...
1•petethomas•6m ago•0 comments

A Meta AI security researcher said an OpenClaw agent ran amok on her inbox

https://techcrunch.com/2026/02/23/a-meta-ai-security-researcher-said-an-openclaw-agent-ran-amok-o...
1•tobr•7m ago•0 comments

Show HN: Microgpt-ts – Full GPT in 500 lines of TypeScript, zero dependencies

https://microgpt-ts.vercel.app/
2•sdubois•7m ago•0 comments

Show HN: Maude the Unicorn Slayer – Disrupting Series-B startups in 16-bit

https://maudetheunicornslayer.com/
1•artlessbfa•7m ago•1 comments

Show HN: Scry – Test migrations against production scale copy of your DB

https://www.scrydata.com/blog/the-postmortem-we-never-wrote/
1•gmcquillan•8m ago•1 comments

Show HN: Tabularis – DB GUI where drivers are JSON-RPC executables

1•debba•10m ago•0 comments

Tesla's Europe problem keeps getting worse

https://www.cnbc.com/2026/02/24/tesla-car-sales-elon-musk-europe-autos-trump-evs.html
1•Betelbuddy•10m ago•0 comments

Claude AI Agents Built a C Compiler:What It Means for the Future of AI Coding

https://manojgopanapalli.substack.com/p/sixteen-claude-ai-agents-built-a
1•thecontentboy•11m ago•0 comments

Show HN: Harp – Offline, Org-Mode Based Personal Health Records Application

https://docs.lepisma.xyz/harp/
1•lepisma•11m ago•0 comments

Show HN: I analyze your GitHub code and generate a developer personality card

https://howyoucode.dev
1•marcelglaeser•12m ago•1 comments

The long tail of niche AI apps on demand

https://www.djmurphy.net/blog/apps-on-demand/
2•sollewitt•12m ago•0 comments

Show HN: AI-Nexus – Unified Rule Manager for Claude Code, Cursor, and Codex

https://github.com/JSK9999/ai-nexus
1•suntrix3•13m ago•1 comments

Conduit AI – AI voice agent that answers missed calls for service businesses

https://www.conduitai.io/
1•wpbluiss•14m ago•1 comments