Open APIs Are Over
After two decades of thriving on open APIs, major software platforms are erecting barriers. Salesforce limited Slack’s API calls, Datadog disabled a rival observability startup, and Epic faces a lawsuit over restricting patient‑record access. The acceleration of AI‑driven development enables incumbents to expand their product suites and defend market share by tightening data portability. This shift signals a new normal where platform walls replace the previously open ecosystem.
Trajectory
The post argues that the software era has moved from merely adopting tools to mastering the specific paths—called trajectories—through which those tools are used. By tracking how users interact with CRM, IDEs, or AI assistants, companies can automate processes, refine...
Effective Strategy Demands Painful Trade‑offs and Clear Ownership
strategy should hurt. if you aren't making painful trade-offs, you haven't prioritized. • one goal, one owner • shared ownership = no ownership molly graham on @lennysan: https://t.co/RW7eAWk6kI
Meta Pays $2B for $100M ARR Startup
Meta is acquiring Manus for $2 billion. Manus disclosed $100 million in ARR achieved in eight months, 147 trillion tokens processed since launch. Can we use those figures to explain the acquisition price?
Gross Profit per Token
Meta announced a $2 billion acquisition of Singapore‑based AI agent firm Manus, which reported $100 million annual recurring revenue and 147 trillion tokens processed since its March 2025 launch. The deal underscores the relevance of gross profit per token as a valuation lens,...
Motive's S‑1 Shows Growth Outpacing Samsara's IPO
Motive, the AI-powered fleet management company formerly known as KeepTruckin, filed their S-1 this week. The company has grown tremendously & in this post, we compare it to Samsara at IPO :
2025 Predictions: Strong IPO Surge, Google Leads AI
Every year I make a list of predictions & score the previous year’s. You can find my 10 Predictions for 2026 here. 2025 was a good year : I scored 7.85 out of 10. 1. The IPO market rips. Score...
AI Agents Will Outprice Human Labor by 2026
11 Predictions for 2026 Every year I make a list of predictions & score last year’s predictions. 2025 was a good year : I scored 7.85 out of 10. Here are my predictions for 2026 : 1. Businesses pay more for AI agents...
12 Predictions for 2026
Tom Tunguz outlines twelve 2026 forecasts, from AI agents commanding higher fees than human workers to a record‑breaking liquidity wave driven by IPOs from SpaceX, OpenAI and others. He predicts vector databases becoming essential AI infrastructure and agents autonomously handling...
Proprietary AI Keeps Pricing Power Despite Cheaper Open‑Source Alternatives
Despite open-source AI models being 10-100x cheaper, proprietary providers haven’t lost pricing power. OpenRouter’s data reveals a market splitting in two. Over the last year, open-source models’ market share has remained stable around 22-25%, briefly spiking to 35% during the explosive...
The Bifurcation in the AI Market
The AI market is splitting into two distinct camps: open‑source models, which remain roughly 22‑25% of usage despite being 10‑100× cheaper, and proprietary providers that retain pricing power for high‑value workloads. Open‑source share held steady, briefly spiking to 35% during...
The Average Founder Ages 6 Months Each Year
The median age of founders raising venture capital is climbing about six months each year, a shift driven by three forces: the surge in AI‑focused funding, the growing dominance of B2B models, and a media narrative that over‑highlights youthful, consumer‑centric...
The Two Context Databases Powering Enterprise AI
Enterprises are adopting two distinct context databases—operational and analytical—to serve as the new system of record for AI agents. Operational databases capture procedural knowledge and trade secrets, while analytical databases encode metric definitions and reasoning logic. The article emphasizes that...
AI Saves Workers Hours, Yet Vendors Capture Only 5%
A new study from OpenAI shows AI saves the average white-collar worker 54 minutes per day. Where does all that value go? BLS data shows median weekly earnings for full-time workers hit $1,165 in Q3 2024, or $60,580 annually. Across 2,080...
The AI Value Gap : Where Does the $7,000 Per Seat Go?
A new OpenAI study finds AI saves white‑collar workers 54 minutes daily, translating to roughly $7,282 of recovered productivity per seat each year. Current SaaS pricing—ChatGPT Plus, Microsoft 365 Copilot, Google Workspace AI, and Gamma—captures only 3‑5% of that value,...
IBM's $11.1B Confluent Deal Signals Streaming Consolidation
Streaming is the next category to consolidate within the modern data stack. IBM announced its intent to acquire Confluent. The deal values Confluent at $11.1 billion, or 10.0x LTM revenue. Confluent commands more than 40% of the Fortune 500 as customers...
Streaming Comes Into the Fold
IBM announced an $11.1 billion acquisition of Confluent, valuing the streaming‑platform provider at 10 × its last‑twelve‑month revenue. Confluent, the commercial arm behind Apache Kafka, serves more than 40 % of Fortune 500 firms and posted Q3 revenue of $298.5 million, up 19.3 % YoY, with...
AI Lets Engineers Multiply Output, Reclaim Leisure Time
Since I watched software engineers using AI, I've become jealous. I've seen them assign 15-20 coding tasks to an AI, play foosball, grab coffee, then return to evaluate the agent's work. That engineer has paralleled her time by 10x to...
From 10 to 31 Tasks Daily : The Agent Inflection Point
An author integrated Google Gemini CLI with Asana, linking AI agents to email, calendar, and CRM, enabling them to execute tasks directly within his workflow. By delegating to agents, he increased his daily output from 10 to 31 tasks, handling...
Cheap Frontier Models Challenge Trillion-Dollar Compute Race
While OpenAI signed $1.15 trillion in compute contracts through 2035, DeepSeek trained a frontier model for $6 million. This was 2025’s central question : are we building on bedrock or quicksand? The top 10 posts of 2025 examined some of these...
Top 10 Posts of 2025
The 2025 "Top 10 Posts" roundup highlights a paradox in AI: OpenAI pledged $1.15 trillion in compute contracts through 2035 while DeepSeek demonstrated a 90%+ drop in training costs with its V3 and R1 models. Nvidia’s $110 billion vendor‑financing scheme dwarfs the...
Six‑Week AI Shift Reduces SDRs Tenfold
I remember hosting a dinner of sales leaders to talk about AI. I asked them what will your CRM do for you in the future? Nearly unanimously came the reply : enable salespeople to spend most of their time with customers. Listening...
AI Cuts SDR Team Tenfold in Six Weeks
I remember hosting a dinner of sales leaders to talk about AI. I asked them what will your CRM do for you in the future? Nearly unanimously came the reply : enable salespeople to spend most of their time with...
Rethink GTM Pricing, Hiring, Compensation for AI Scale
Office Hours: GTM Leadership in the AI Era: Pricing, Hiring, and Compensation Thursday, December 11 5:30–7:30 PM PDT As AI-native companies scale, GTM leaders are being forced to rethink long-standing assumptions about pricing, hiring, incentives, and how revenue organizations should operate in a...
Rethinking GTM Pricing, Hiring, Compensation for AI Scale
Office Hours: GTM Leadership in the AI Era: Pricing, Hiring, and Compensation Thursday, December 11 5:30–7:30 PM PDT As AI-native companies scale, GTM leaders are being forced to rethink long-standing assumptions about pricing, hiring, incentives, and how revenue organizations should...
Private Equity : The New Distribution Channel for AI Startups
The post argues that private equity (PE) firms have become the primary distribution channel for AI startups, thanks to the rapid growth of PE‑owned companies—now outnumbering public firms 4:1 in the US. Because PE portfolios consist mainly of mid‑market firms...
AI Market Sizzles: Rapid Change Fuels Excitement
The AI market today is bacon in a hot skillet. Everything is sizzling, moving, & changing at an incredible pace. We’re all watching it closely. Market share is fluid because no one yet knows what AI can do & the...
The Bacon & the Skillet: When Does the AI Market Congeal?
The post likens the rapidly evolving AI market to sizzling bacon, emphasizing that fluid market share and high switching costs are driven by continual model improvements like Nvidia chips and Gemini 3. It argues that while the current phase rewards speed...
Nvidia Earnings and Gemini Tweet Reshape AI Narrative
Two revelations this week have shaken the narrative in AI : Nvidia’s earnings & this tweet about Gemini. The AI industry spent 2025 convinced that pre-training scaling laws had hit a wall . Models weren’t improving just from adding more...
The Scaling Wall Was A Mirage
The post argues that the supposed "scaling wall" in AI pre‑training has been disproven by recent developments: Gemini 3 achieved dramatic performance gains without increasing parameters, showing that algorithmic and compute improvements still drive scaling laws, and Nvidia's record earnings and...
Production AI Teams Favor Open‑Source Models and Database‑Backed Agents
70% of production AI teams use open source models. 72.5% connect agents to databases, not chat interfaces. This is what 375 technical builders actually ship - & it looks nothing like Twitter AI. 70% of teams use open source models...
Microsoft Pivots to Agent‑centric Infrastructure Business
"Our business, which today is an end-user tools business, will become essentially an infrastructure business in support of agents doing work." — Satya Nadella Microsoft talking about their view where the value accrues : agent infrastructure https://t.co/1TaxJpOeCB
Distilling Tool-Calling Skills From Claude to Smaller Model
Ten months ago, DeepSeek collapsed AI training costs by 90% using distillation - transferring knowledge from larger models to smaller ones at a fraction of the cost. Distillation works like a tutor training a student : a large model teaches a...
DeepSeek Cuts AI Training Costs 90% via Distillation
Ten months ago, DeepSeek collapsed AI training costs by 90% using distillation - transferring knowledge from larger models to smaller ones at a fraction of the cost. Distillation works like a tutor training a student : a large model teaches...
Model Firms Risk Winner's Curse without Unique Data
If you're a model company, you may have a winner's curse. You did all the hard work... except it's kind of like one copy away from being commoditized. And then the person who has the data for grounding can go...
Teaching Local Models to Call Tools Like Claude
The author reports a proof‑of‑concept that distills tool‑calling behavior from Claude into a local 20B GPT‑OSS model by logging Claude’s tool selections, curating the data with SemDeDup and CaR, and iteratively refining prompts using DSPy and GEPA. Over three optimization...
Datadog's Platform Shift Drives 25% Quarterly Growth
Datadog is becoming a platform company, & its Q3 2025 results underscore how successful this transition is. If nothing else, the consistency around 25% growth for the last 12 quarters exemplifies this point. Net dollar retention underpins this growth, combined...
AI Promotions Spark Unprecedented Unsubscribe Surge, Exposing Reader Fear
I wrote a post about AI promotions. It triggered the highest unsubscribe rate in this blog's history. A 4-sigma event - expected once every 700 years. The data reveals what readers feared 🧵
Are We Being Railroaded by AI?
The post compares AI infrastructure spending to historic U.S. investment projects, showing AI at 1.6% of GDP in 2024—far below World War II, the New Deal, and the railroad boom. It highlights current corporate outlays (Microsoft, Google, Meta) and projects OpenAI’s...
OpenAI Plans $1.15 Trillion Infrastructure Investment by 2035
OpenAI has committed to spending $1.15 trillion on hardware & cloud infrastructure between 2025 & 2035. 1 The spending breaks down across seven major vendors: Broadcom ($350B), Oracle ($300B), Microsoft ($250B), Nvidia ($100B), AMD ($90B), Amazon AWS ($38B), & CoreWeave...
Cloud‑migrated DuckDB Slashes Podcast Processing Costs
I sleep better knowing my agents work through the night. Less work for me in the morning. My podcast processor transcribes & analyzes conversations. I started on my laptop, needed a little database to collect podcast data & metadata, & booted...
Night‑Shift Agents Let Me Sleep Better
I sleep better knowing my agents work through the night. Less work for me in the morning. My podcast processor transcribes & analyzes conversations. I started on my laptop, needed a little database to collect podcast data & metadata, &...
$555B of Cloud Spend
Microsoft and Google reported surging AI-driven cloud demand, with combined remaining performance obligations and capex commitments swelling to roughly $555 billion as both ramp data-center builds and GPU capacity. Microsoft’s commercial RPO topped nearly $400 billion and Google’s backlog hit...
Double Speed, Ten Percent Cost: Game-Changing Efficiency
Twice as fast & 10% of the cost! https://t.co/Z5m1T8nleq
Public SaaS Stagnates; Private AI Unicorns Lead Growth
As the world follows every whisper & rumor from AI, what has happened to public SaaS? The answer is not much! All the fun is in the private markets. Multiples haven’t moved outside of a narrow band since the post-Covid crash in...
Product‑Market Fit Is Ongoing, Not a One‑Time Milestone
There’s this notion that product market fit is this binary condition that once you pass the gate, you’re done. And that’s no longer the case. Product market fit is like United Miles. You continue to fly to maintain 1K. And...
Product-Market Fit Is No Longer Static
Venture investor Tomasz Tunguz argues product-market fit (PMF) is no longer a one-time milestone but a continuous condition because rapid advances in AI are constantly reshaping customer needs and technical possibilities. He points to firms that lost PMF between 2021–2024,...
Scaling AI: Strategies for Building Intelligent Systems
How do you build AI systems that scale? On Nov 12 in SF, I’ll host Calvin French-Owen (Co-founder & former CTO @Segment, led @OpenAI's Codex launch) for a Theory Ventures Office Hours conversation on scaling intelligent systems. We’ll dig into: ⚙️ Building...
From Columns to Rewards: Automating the Two Pillars That Drive Modern AI
Former Google engineers automated feature engineering with AutoML—turning manual spreadsheet column tinkering into processes that added tens to hundreds of millions in revenue—and the author argues reinforcement learning (RL) is now at that same inflection point. Today RL can generate...
Choose the Right Location for Your AI Agents
It’s 9:30 AM. Do you know where your agent is? As we enter the era of agentic AI, this is an increasingly important question. ChatGPT launched the consumer use of AI, compressing all of human knowledge into a single model....