The Futurist

Open-source memory layer that teaches AI agents to remember and learn

“Persistent memory is the missing piece between 'AI assistant' and 'AI colleague.' Stash's self-correction and failure pattern recognition are early implementations of what agents will need to become genuinely reliable over long time horizons.”

A true 1-bit 8B LLM that fits in 1.15 GB — runs on your iPhone

“The trajectory here is what matters: 1-bit models are getting faster to train and competitive faster than expected. When custom Apple Neural Engine kernels land for BitNet-style weights, we'll see 200+ tokens/sec on a phone. Bonsai-8B is the proof-of-concept that makes that future feel real.”

WiFi-based AI pose detection and vitals monitoring — no cameras

“Camera-free sensing is foundational infrastructure for a world where AI monitors physical spaces without the privacy baggage of video. Elder care, physical rehabilitation, smart home automation — all of these become viable in privacy-sensitive contexts once you remove the camera. At $9 per node, mass deployment is economically possible for the first time.”

Idle Macs become a decentralized AI inference network — 70% cheaper

“This is Napster for AI compute — and I mean that as a compliment. If Darkbloom cracks the reliability and routing problem, it could force AWS and GCP to dramatically cut inference prices or lose the long tail of developers entirely. The decentralized compute flywheel is finally legible.”

Deploy and distribute AI apps and MCP servers from one platform

“The first company to become the App Store for MCP servers will capture enormous value in the agentic AI economy. Alpic is early to a market that will be worth billions. The open Skybridge standard is a smart move to avoid the walled-garden trap. If they nail developer experience before the big platforms wake up, they could define the category.”

Open-source micro VMs for running AI agents, browser tasks, and computer-use workflows

“Compute sandboxing is becoming AI's next infrastructure layer — the thing every agentic system needs but nobody wants to build twice. Open-source here is the right call; just as databases and caches became infrastructure commodities, execution sandboxes will too.”

The GitHub of machine learning — models, datasets, and Spaces

“Hugging Face is the open-source counterweight to closed AI labs. They are democratizing access to AI in a way that matters for the entire industry.”

Build with Claude API — prompt engineering, evaluation, and deployment

“Anthropic is building the developer platform, not just the model. Console + Claude Code + Agent SDK — they want developers building on Claude, not just chatting with it.”

Containerize anything — the standard for packaging and deploying apps

“Containers are the universal packaging format for software. AI agents, ML models, microservices — everything ships in containers. Docker is infrastructure.”

Run open-source AI models with one API call

“Replicate is making open-source AI as easy to use as closed APIs. That is the right mission at the right time.”

Fastest LLM inference — custom silicon for instant responses

“Custom silicon for LLMs is the right long-term bet. GPUs are general-purpose. Groq is purpose-built. As open-source models match GPT quality, Groq becomes the default inference layer.”

Fast inference for open-source LLMs at low cost

“Together is betting that the future is open-source models. As Llama and Mistral improve, inference providers like Together become the AWS of AI.”

GPT API, Assistants, fine-tuning, and the playground

“OpenAI has the largest ecosystem of developers and integrations. Even if other models catch up, the platform moat is real.”

Deploy app servers close to your users globally

“Fly.io is the answer for workloads that don't fit the serverless model. As AI inference goes local-first, having servers in 30+ regions matters.”

Edge computing at 300+ locations worldwide

“Cloudflare is building the programmable internet. Workers + D1 + R2 + AI = a complete platform that runs at the edge. They're quietly becoming the default infrastructure layer.”

Serverless Redis and Kafka — per-request pricing

“Upstash is doing for Redis what Neon did for Postgres — making it serverless-native. The QStash message queue is an underrated piece of the puzzle.”

Payment infrastructure with AI-powered fraud detection and revenue tools

“Stripe is quietly becoming the financial infrastructure for the internet. Atlas, Treasury, Issuing — they're building the operating system for internet businesses.”

Open-source Firebase alternative with Postgres, auth, and AI

“Supabase proves that open-source alternatives can match and exceed proprietary platforms. They're building Firebase, but better, and you can self-host if you want.”

Frontend cloud platform — deploy Next.js and more with zero config

“Vercel is building the cloud for the AI era — AI Gateway, Workflow DevKit, Edge Functions. They're not just a hosting platform, they're the application platform.”

Serverless Postgres with branching and instant scaling

“Neon is making Postgres behave like a serverless primitive. The branching model will become standard — in 3 years, we'll wonder how we ever managed databases without it.”

Fast serving framework for LLMs

“Constrained decoding and structured generation are the future of reliable LLM outputs. SGLang leads here.”

Run AI models on Cloudflare's network

“Edge AI inference will be standard for latency-sensitive applications. Cloudflare's network provides unique distribution.”

Fully managed foundation model service

“AWS's distribution advantage means Bedrock will be how most enterprises consume AI models.”

High-throughput LLM serving engine

“Self-hosted inference will remain important for latency, cost, and privacy. vLLM is the infrastructure layer.”

Sandboxed cloud environments for AI agents

“Safe code execution for AI agents is critical infrastructure. E2B is building the sandbox layer that every agent needs.”

Hugging Face text generation inference

“Hugging Face's ecosystem play — models, datasets, spaces, inference — creates a compelling end-to-end platform.”

Fastest inference for open and custom models

“The inference provider market is heating up. Fireworks' focus on reliability and speed builds trust.”

Serverless cloud for AI and data

“Serverless GPU is the future of AI compute. Modal's developer experience is setting the standard.”

Open-source self-hosting platform

“The self-hosting movement is growing. Coolify makes it accessible to developers who don't want to be sysadmins.”

Remote container builds for CI

“Remote build infrastructure will become standard. Local or CI builds on underpowered machines make no sense.”

Serverless GPU inference

“Specialized GPU inference for media generation is a growing market. fal.ai's speed creates a real differentiator.”

Observability for serverless

“Observability built into the platform (Cloudflare, Vercel) rather than bolted on is the right direction.”

Self-hosted monitoring tool

“Self-hosted monitoring that looks great and works reliably. Open-source infrastructure monitoring is mature.”

Serverless JavaScript at the edge

“Deno Deploy proves the tight runtime + platform integration creates the best DX. Bun will follow this pattern.”

Google Cloud's ML platform

“Google's AI infrastructure advantage (TPUs, models, data) makes Vertex the dark horse enterprise AI platform.”

Build modern full-stack apps on AWS

“SST proves AWS can have great DX. Ion's Pulumi-based multi-cloud approach positions it for the future.”

Log management and observability

“Query-on-read architecture is the future of log management. Axiom's approach makes unlimited logging practical.”

Deploy apps and databases instantly

“Railway is building the best developer experience in cloud hosting. They understand what developers actually want.”

Scalable AI compute platform

“Ray is becoming the distributed computing standard for AI. Anyscale manages the hard parts.”

Observability framework for cloud-native software

“OpenTelemetry will be to observability what Kubernetes is to orchestration — the universal standard.”

Microsoft's AI services platform

“Microsoft's investment in OpenAI and enterprise distribution makes Azure AI the default for Fortune 500 AI adoption.”

Infrastructure as code in any programming language

“AI can write TypeScript better than HCL. Pulumi's approach is more natural for the AI-assisted future.”

GPU-optimized AI software catalog

“NVIDIA's software ecosystem (CUDA, TensorRT, Triton) is as important as their hardware. NGC is the distribution layer.”

Deploy app servers close to your users

“Apps running close to users is the future. Fly.io's Machines API enables new categories of distributed applications.”

Observability for distributed systems

“As systems grow more complex, observability tools that surface problems automatically become essential. Honeycomb leads here.”

Cloud-native reverse proxy and load balancer

“Service mesh and API gateway convergence is happening. Traefik is well-positioned at that intersection.”

The ultimate server with automatic HTTPS

“Caddy proves that web servers can be simple. Automatic HTTPS should have always been the default.”

Serverless compute on AWS

“Serverless is the default compute model. Lambda's ecosystem and AWS integration ensure its dominance.”

Web development platform for the modern web

“Vercel's framework-first approach and AI features are winning developer mindshare. Netlify needs a new differentiator.”

Infrastructure as code for any cloud

“Infrastructure as code is table stakes. Terraform's provider ecosystem is its moat and it keeps growing.”

Container orchestration at scale

“The API model Kubernetes established is becoming the universal infrastructure abstraction layer.”

Open-source observability and dashboarding

“The LGTM stack (Loki, Grafana, Tempo, Mimir) is a credible open-source alternative to the entire Datadog suite.”

Google's app development platform

“Firebase defined BaaS but Supabase's open-source, Postgres-based approach is the future.”

Open-source monitoring and alerting

“Prometheus + Grafana is the open-source observability stack. Mimir extends it to global scale.”

Cloud infrastructure for developers

“Squeezed between hyper-scalers above and serverless platforms below. The mid-market IaaS space is shrinking.”

Cloud monitoring and security platform

“Datadog's land-and-expand across security, CI, and database monitoring makes them the observability platform to beat.”

Affordable European cloud hosting

“As cloud bills become a board-level concern, Hetzner's value proposition becomes more compelling every quarter.”