热门技术推文 - 2026年4月29日

2026年4月29日科技每日简报

Today's top tech conversations are led by @FirstSquawk, whose post about 'OPENAI MISSES USER AND REVENUE...' garnered the highest engagement. Key themes trending across the top stories include agent, agents, openai, production, claude. The community is actively discussing recent developments in AI, engineering practices, and startup strategies.

1. FirstSquawk (Group Score: 501.7 | Individual: 44.5)

Cluster: 19 tweets | Engagement: 761 (Avg: 93) | Type: Tech

OPENAI MISSES USER AND REVENUE TARGETS, RAISING CONCERNS OVER DATA CENTER SPENDING SUSTAINABILITY: WSJ

See 18 related tweets

@WSJ: Exclusive: OpenAI recently missed its own targets for new users and revenue, stumbles that have rais...
@edzitron: I don’t think people realize the ramifications of OpenAI’s revenue growth slowing. Oracle is buildin...
@edzitron: Oh hey look it’s the thing I’ve been warning about https://t.co/S4bUgRUF52\n\nQT @berber_jin1: NEW: ...
@edzitron: RT @edzitron: I don’t think people realize the ramifications of OpenAI’s revenue growth slowing. Ora...
@business: OpenAI recently failed to meet its own goals for new user acquisition and sales, fueling internal co...

2. business (Group Score: 223.1 | Individual: 27.5)

Cluster: 11 tweets | Engagement: 189 (Avg: 68) | Type: Tech

China has sought for years to exert influence over business deals beyond its home turf.

Still, its decision to press Meta to unwind a $2 billion acquisition of AI startup Manus marks a step unlike anything it’s tried before. https://t.co/tqRzFox6HN

See 10 related tweets

@WSJopinion: If you’re an artificial-intelligence startup or investor in China, you can check in but you can’t ch...
@SarithaRai: RT @business: China’s decision to press Meta to unwind a $2 billion acquisition of AI startup Manus ...
@WSJ: Exclusive: Meta Platforms is preparing to have to unwind its acquisition of the artificial-intellige...
@StockSavvyShay: $META is preparing to unwind its acquisition of AI startup Manus after China reportedly blocked the ...
@StockMKTNewz: Meta Platforms $META is preparing to have to unwind its acquisition of the AI startup Manus after Ch...

3. tetsuoai (Group Score: 201.4 | Individual: 62.5)

Cluster: 5 tweets | Engagement: 53164 (Avg: 1970) | Type: Tech

RT @elonmusk: Scam Altman and Greg Stockman stole a charity. Full stop.

Greg got tens of billions of stock for himself and Scam got dozens of OpenAI side deals with a piece of the action for himself, Y Combinator style. After this lawsuit, Scam will also be awarded tens of billions in stock directly.

The fundamental question is simply this:

Do you want to set legal precedent in the United States that it is ok to loot a charity? If so, you undermine all charitable giving in the United States forever.

I could have started OpenAI as a for-profit corporation. Instead, I started it, funded it, recruited critical talent and taught them everything I know about how to make a startup successful FOR THE PUBLIC GOOD.

Then they stole the charity.

See 4 related tweets

@MarioNawfal: 🚨 ELON IS IN THE COURTROOM

Day 2 of the OpenAI trial. He showed up in person.

His argument: Altman...

@cb_doge: Elon Musk is NOT fighting OpenAI for money. The proceeds of any legal victory in the OpenAI case wil...
@Grady_Booch: I stand with @elonmusk\n\nQT @elonmusk: Scam Altman and Greg Stockman stole a charity. Full stop.

...

@XFreeze: How is it not pure evil to flip from a non-profit to a for-profit, and then refuse to give equity to...

4. cryptopunk7213 (Group Score: 192.3 | Individual: 33.4)

Cluster: 10 tweets | Engagement: 133 (Avg: 416) | Type: Tech

amazon $AMZN is by far the most underrated ai company in the world right now and nobody sees it:

owns 20% of anthropic. owns major stake in openai (investing $35B)

primary provider of claude, chatgpt and any top ai model via amazon bedrock

EVERY government organisation uses amazon to manage and serve ai models

AWS revenue due to ai is through the fucking roof.

amazon trainium ai chips are sold-out including their new chip that hasn’t released yet. demand is at ath

anthropic committed to using 1-2M amazon chips to train claude

amazon robotics division is one of the largest in the world. data can be used to train top robotics models

spending the most on ai capex ($200B) this year alone. ceo Jassy says demand has already justified the spend. if they’re right they’ll be right in a huge way.

my bet is tmrw they announce a major partnership with openai that includes ability to distribute multiple enterprise versions of chatgpt to amazons fortune 500 clients.

amazon has silently become the #1 distributor of the worlds best models

that alone is worth trillions.\n\nQT @ajassy: Very interesting announcement from OpenAI this morning. We’re excited to make OpenAI's models available directly to customers on Bedrock in the coming weeks, alongside the upcoming Stateful Runtime Environment. With this, builders will have even more choice to pick the right model for the right job. More details at our AWS event in San Francisco tomorrow.

See 9 related tweets

@cryptopunk7213: fantastic interview with sam altman and amazon AWS ceo matt garman on todays partnership announcemen...
@StockMKTNewz: Amazon $AMZN just said it will make OpenAI's AI models available to its customers after Microsoft dr...
@cryptopunk7213: massive earnings day tmrw - google, amazon, microsoft, meta and apple are spending $600B+ on AI this...
@zephyr_z9: RT @cryptopunk7213: fantastic interview with sam altman and amazon AWS ceo matt garman on todays par...
@StockSavvyShay: $AMZN AWS launched Amazon Connect Decisions and Connect Talent which is a new AI agent tools for rec...

5. dkundel (Group Score: 179.8 | Individual: 48.3)

Cluster: 7 tweets | Engagement: 1342 (Avg: 126) | Type: Tech

RT @thsottiaux: Don't just reset Codex rate limits for fun, it costs money. Don't just reset Codex rate limits for fun, it costs money.

... but the vibes are good ...

I have reset Codex rate limits for ALL paid plans to celebrate a good week and allow everyone to build more with GPT-5.5. Enjoy

See 6 related tweets

@reach_vb: Rate limits RESET! 🔥

Vibes are through the roof with GPT 5.5 in Codex, now you can enjoy codex even...

@romainhuet: Wherever you are in the world, whether it’s morning, afternoon, or night…

Rate limits have been res...

@TheRealAdamG: He givith https://t.co/1KmpEzn73J\n\nQT @thsottiaux: Don't just reset Codex rate limits for fun, it ...
@scaling01: Anthropic is getting capacity mogged\n\nQT @thsottiaux: Don't just reset Codex rate limits for fun, ...
@Dimillian: Hello EU it’s reset time!\n\nQT @thsottiaux: Don't just reset Codex rate limits for fun, it costs mo...

6. codewithimanshu (Group Score: 179.1 | Individual: 34.5)

Cluster: 6 tweets | Engagement: 872 (Avg: 437) | Type: Tech

RT @codewithimanshu: Anthropic's Claude Ai Agents Team just Educated how to build production AI agents in under 30 mins.

For Free. From the engineers who built the stack.

CANCEL Your Weekend Plans, and Learn to Build AI Agents Today.

Bookmark it. Watch it. Build your first production agent this weekend.

$5,000/month.$ 7,000/month. $12,000/month.

People are building agents for clients and charging $$$ as Beginners. You're still stuck in the thinking about AI phase.

This video fixes that tonight.

Follow @codewithimanshu for more high-signal content that actually moves your AI engineering career forward. ↓

Ivan Nardini runs Developer Relations for AI at Google Cloud. He just gave away the entire production agent stack in 30 minutes.

This is the talk that separates people deploying AI agents that actually scale from people whose agents break the moment they leave localhost.

Here's everything inside.

I break down a production AI video like this every week. Follow @codewithimanshu.

↓ The 4-part agent stack that actually scales.

Most devs are duct-taping frameworks together and calling it an "AI agent."

Ivan lays out the real stack:

Agent Development Kit (ADK): open-source, code-first framework for building, evaluating, and deploying agents. Supports Claude models through Vertex AI directly.

Model Context Protocol (MCP): lets your agent talk to any tool or data source with one standard. Vertex AI Agent Engine: managed platform for deploying, monitoring, and scaling agents in production. No DevOps headaches.

Agent-to-Agent Protocol: open protocol so agents built on different frameworks can actually work together.

This is the stack replacing every hacky agent setup in production right now. Full MCP + Claude breakdowns drop weekly on @codewithimanshu.

↓ Building your first real agent. Ivan builds a birthday planner agent live.

LLM Agent class. Name it. Define instructions. Pick the model.

He uses Claude 3.7 Sonnet. You could use Opus 4.7 for better reasoning.

Full agent built in minutes. Not weeks.

Watch the build once and you'll never structure an agent the wrong way again.

I post agent architectures people pay $500 courses to learn. @codewithimanshu.

↓ Multi-agent systems without the chaos.

Single agents are easy. Multi-agent systems are where 99% of builders fail.

Ivan extends the birthday planner by: Adding a calendar service through MCP tools Creating an orchestrator agent to route requests between agents Handling state and context across agent handoffs

This is production multi-agent architecture. Clean. Scalable. Debuggable. Most tutorials hand-wave this part. This one shows you every step.

Multi-agent orchestration content drops weekly on @codewithimanshu.

↓ Deployment without the DevOps nightmare. This is where most AI projects die.

You build a cool agent locally. It works. You try to deploy it. Everything breaks. Vertex AI Agent Engine fixes this:

Minimal code deployment Automatic monitoring of latency, CPU, and memory Built-in observability and logging No infrastructure setup needed

You provide config and requirements. The platform handles the rest. This is how agents actually get to production.

Deployment guides for Claude agents post every week. @codewithimanshu.

↓ Agent-to-Agent Protocol: the future nobody's talking about. Most people don't know this exists yet.

The A2A Protocol lets agents built in different frameworks communicate seamlessly.

Your Claude agent. My LangChain agent. Someone else's CrewAI agent. All talking to each other. All solving parts of the same problem. All without custom integration code.

This is the infrastructure layer of the coming AI economy. Getting in early on A2A Protocol is like getting in early on HTTP in 1995.

A2A deep dive coming soon. @codewithimanshu.

↓ 30 minutes from the team shipping this in production.

You'll learn more from this than from 6 months of YouTube tutorials made by people who've never deployed an agent past localhost.

People who watch this understand production AI agents at the architect level. People who skip it keep hacking together frameworks that break every time an API updates.

Save the video. Watch it tonight. Build a real agent this weekend.

Follow @codewithimanshu for more high-signal content that actually moves your AI engineering career forward.

See 5 related tweets

@codewithimanshu: Stanford just dropped a free 17-video course that teaches you to build Claude. From scratch.

There ...

@codewithimanshu: RT @codewithimanshu: Anthropic's Claude Code team just teaches how to automate your entire engineeri...
@codewithimanshu: RT @codewithimanshu: Stanford just dropped a free 17-video course that teaches you to build Claude. ...
@codewithimanshu: RT @codewithimanshu: Anthropic's Applied AI team just dropped a 25-minute talk on how to actually us...
@AnandButani: 🔥 10 New AI Tools Worth Watching

https://t.co/6fyX5UeL0R: general-purpose AI agent that runs long,...

7. theinformation (Group Score: 177.2 | Individual: 27.4)

Cluster: 10 tweets | Engagement: 0 (Avg: 7) | Type: Tech

There’s still remaining friction between OpenAI and Microsoft after it averted a legal war over AWS.

"We learned that under the new deal... Microsoft still isn't entitled to all of the intellectual property that OpenAI would get from doing an acquisition." — @aaronpholmes https://t.co/3VWnsW95SZ

See 9 related tweets

@business: Amazon said it will finally be able to make OpenAI's models available to its own customers after Mic...
@chiefofautism: openai is VERY defocused company and derailing from their original path

i would be very mad as inve...

@Techmeme: Amazon and OpenAI announce an expanded deal that will make OpenAI's models available from AWS, a day...
@theinformation: Microsoft and OpenAI rewrote their partnership after a clash over OpenAI’s move to sell AI through A...
@StockSavvyShay: OpenAI’s new $MSFT agreement reportedly doesn't cover technology it gains through acquisitions givin...

8. clairevo (Group Score: 161.5 | Individual: 48.0)

Cluster: 5 tweets | Engagement: 634 (Avg: 113) | Type: Tech

Built a realtime phonics + reading practice app that adapts to the reader's mistakes and reinforces sound-it-out.A little laggy because I'm running locally but so excited about this for my early reader! https://t.co/n34nQfDYYT\n\nQT @OpenAIDevs: You can build interactive applications with gpt-realtime-1.5, so users can control app state more naturally with voice.

Hi Chappy 👋 https://t.co/mh1O8ZBzIY

See 4 related tweets

@PaulSolt: RT @OpenAIDevs: You can build interactive applications with gpt-realtime-1.5, so users can control a...
@romainhuet: The Realtime API is fast.

This demo by @jxnlco is running at real speed! No computer use, just real...

@VaibhavSisinty: The future of interface is def voice

This seems to be the closest I have seen to be usable.

Laten...

@EricNewcomer: Hear from @juberti at Cerebral Valley Voice next week!\n\nQT @OpenAIDevs: You can build interactive ...

9. chatgpt21 (Group Score: 159.0 | Individual: 30.2)

Cluster: 6 tweets | Engagement: 48 (Avg: 180) | Type: Tech

Anthropic releases text to cad and now base44 automates sales force pipelines?!\n\nQT @Base44: Introducing: Base44’s new platform migration feature.

If your needs have outgrown your current software, you can now bring your projects from other platforms into Base44 in a single click. This includes your Salesforce pipeline, Shopify catalog or WordPress site. Also whatever you’ve built on Lovable, Bolt, or Replit.

To celebrate the launch, anyone who completes migration by May 5th, 12am ET gets 25 free credits added to their account!

See 5 related tweets

@dr_cintas: Moving from one platform to another usually means months of migration work.

Base44 just made it a s...

@kimmonismus: OpenAI and Anthropic are racing to build the smartest agents.

Base44 is building the platform thos...

@svpino: This is not only about starting a new project using AI.

You can now import an existing project ...

@shiri_shh: babe wake up.

@Base44 just dropped one-click MIGRATION from WordPress, Lovable, Salesforce, Shopify...

@cgtwts: POV: you realize you can rebuild your own HubSpot or Shopify and it moves everything in one click ht...

10. ai_explorer25 (Group Score: 155.0 | Individual: 31.8)

Cluster: 6 tweets | Engagement: 122 (Avg: 92) | Type: Tech

DeepSeek is quietly doing what most labs promise pushing top-tier AI into everyone’s hands without the usual barriers.

DeepSeek isn’t just competing on benchmarks anymore, it’s competing on accessibility.

And with ZenMuxAI making it instantly usable, the gap between “cutting-edge” and “actually usable” just got a lot smaller.

Feels less like a launch, more like a shift in how fast builders can move now.\n\nQT @ZenMuxAI: DeepSeek-V4 @deepseek_ai is HERE FREE on @ZenMuxAI ! 🚀
API & Chat are free for a limited time – jump right in!

Two editions, one mission:
🔷 V4-Pro – max performance, built for agentic coding. Real‑world feel: better than Sonnet 4.5, close to Opus 4.6 (non‑thinking).
⚡ V4-Flash – smaller, faster, lower cost. Simple agent tasks? On par with Pro.

🔥 One API, all protocols – OpenAI Chat/Responses + Anthropic Messages. Use it inside Codex, Claude Code, OpenCode, etc.
PAYGO & Builder Plan already supported.

Try for free now → @ZenMuxAI

#DeepSeek #DeepSeekV4 #AI #LLM #VibeCoding #OpenSource #ZenMuxAI

See 5 related tweets

@Parul_Gautam7: Idea → execution just got way faster.

No friction, no getting stuck mid-build.

Just open it up and...

@Origin_AI_01: Free is a strategy, not a feature.

When frontier models drop cost to zero, distribution and develop...

@TheoBuildsAI: The “fast + cheap” version is honestly what caught my attention. Not everything needs max power all ...
@TheoBuildsAI: RT @ZenMuxAI: DeepSeek-V4 @deepseek_ai is HERE FREE on @ZenMuxAI ! 🚀
API & Chat are free for a l...
@RoundtableSpace: A FREE AI CODER FROM CHINA IS STARTING TO COMPETE WITH PAID TOOLS LIKE CHATGPT.

DeepSeek V4 is gain...

11. FireworksAI_HQ (Group Score: 148.5 | Individual: 32.1)

Cluster: 7 tweets | Engagement: 46 (Avg: 71) | Type: Tech

Excited to support @NVIDIA Nemotron 3 Nano Omni, now available on Fireworks.

It's the first open model that handles vision, audio, video, and text in a single inference loop. Built for multimodal sub-agents at scale, with 9× higher throughput than Qwen3 30B. 256K context.

Now available on Fireworks for on-demand deployments. →https://t.co/qQZr5MGhWg

See 6 related tweets

@UnslothAI: NVIDIA releases Nemotron-3-Nano-Omni, a new 30B open multimodal MoE model.

Nemotron-3-Nano-Omni-30B...

@FirstSquawk: NVIDIA IS INTRODUCING THE NEMOTRON 3 NANO OMNI MODEL, WHICH COMBINES VISION, AUDIO, AND LANGUAGE CAP...
@pingToven: JOE NEMOTRONNNNNN\n\nQT @OpenRouter: NVIDIA Nemotron™ 3 Nano Omni is live on OpenRouter.

An open 30...

@LightningAI: Everyone wants to build multimodal agents but most get stuck stitching together separate vision, aud...
@pingToven: RT @OpenRouter: NVIDIA Nemotron™ 3 Nano Omni is live on OpenRouter.

An open 30B-A3B multimodal mode...

12. ai_explorer25 (Group Score: 144.8 | Individual: 34.6)

Cluster: 5 tweets | Engagement: 183 (Avg: 92) | Type: Tech

Curious how it holds up with more detailed inputs\n\nQT @HeyGen: HyperFrames, now an official plugin in Codex

One click install

Codex becomes your end-to-end video workspace

Code it, edit it, render it, ship it

Stay tuned for more HeyGen x @OpenAIDevs integrations this week! https://t.co/Qp3iED5y5p

See 4 related tweets

@Origin_AI_01: Creative tooling is collapsing into one loop.

Idea → code → video → publish.

Whoever owns that loo...

@TheoBuildsAI: I like that it’s not a black box. You can tweak things directly instead of starting over every time....
@Parul_Gautam7: Adding this to my “experiment soon” list 👀\n\nQT @HeyGen: HyperFrames, now an official plugin in Cod...
@TheoBuildsAI: RT @HeyGen: HyperFrames, now an official plugin in Codex

One click install

Codex becomes your end-...

13. Reuters (Group Score: 142.1 | Individual: 18.5)

Cluster: 10 tweets | Engagement: 83 (Avg: 98) | Type: Tech

Google signs classified AI deal with Pentagon, The Information reports https://t.co/4ZBQDp5mfO https://t.co/4ZBQDp5mfO

See 9 related tweets

@nic_carter: Remember when 4000 Google employees threw a fit around running a datacenter for the Pentagon in 2018...
@WSJ: Google agreed to let the Defense Department use its artificial-intelligence tools in classified sett...
@FirstSquawk: The Information reports that Google secured a classified AI contract with the Pentagon amid internal...
@StockSavvyShay: $GOOGL signed a deal to let the Pentagon use its AI models for classified work and “any lawful gover...
@Cointelegraph: 🇺🇸 BIG: Google signs a classified AI deal with the Pentagon, The Information reports. https://t.co/4...

14. chris_j_paxton (Group Score: 141.4 | Individual: 41.6)

Cluster: 4 tweets | Engagement: 302 (Avg: 64) | Type: Tech

RT @asimovinc: We're open-sourcing Asimov v1, a humanoid robot.

With Asimov v1, you can build, train on, and make it your own humanoid robot. It's the first step of building a humanoid labor force for the rest of us.

Asimov v1 is 1.2 m tall, 35 kg, with 25 actuated degrees of freedom. Structural parts machined in 7075 aluminium and 3D-printed in MJF PA12 nylon.

We're releasing the mechanical design and simulation files. Ready for locomotion policy training out of the box.

The BOM is open too. Source everything yourself, or order the DIY Kit. All components, ready to assemble. $499 deposit,$ 15,000 target price. Ships end of summer 2026.

GitHub: https://t.co/kjqkny2oqW Manual: https://t.co/9tjkteOcxO DIY Kit: https://t.co/tzvzNyXQfA

Most humanoid robots are controlled by the companies that build them. Asimov v1 is built for the rest of us. Build it, test it, and share your feedback with the community.

See 3 related tweets

@chris_j_paxton: A hackable, free, open-source humanoid for $15,000.\n\nQT @asimovinc: We're open-sourcing Asimov v1,...
@Baconbrix: my downstairs neighbors are in for a treat over the next couple months\n\nQT @asimovinc: We're open-...
@WesRoth: Asimov open-sourced the mechanical design, simulation files, and complete Bill of Materials (BOM) fo...

15. business (Group Score: 138.8 | Individual: 24.4)

Cluster: 8 tweets | Engagement: 43 (Avg: 68) | Type: Tech

Apple is planning a major overhaul of the built-in photo-editing features for the iPhone, iPad and Mac, leaning heavily on artificial intelligence to better compete with Android devices https://t.co/woEmh0hYeu

See 7 related tweets

@markgurman: NEW: Apple plans AI overhaul for photo editing in iOS 27 and macOS 27, looking to better compete wit...
@markgurman: Development of the features hasn’t gone entirely smoothly, though. The Extend and Reframe tools don’...
@FirstSquawk: APPLE INTENDS TO REVAMP PHOTO EDITING WITH AI IN iOS 27 LATER THIS YEAR, ENHANCING ITS PHOTO CAPABIL...
@StockSavvyShay: $AAPL is planning a major AI photo-editing overhaul across iOS 27, iPadOS 27 and macOS 27 this fall....
@wallstengine: Apple is planning a major AI photo-editing overhaul for iOS 27, iPadOS 27, and macOS 27 this fall, a...

16. ai_explorer25 (Group Score: 138.0 | Individual: 33.7)

Cluster: 6 tweets | Engagement: 1 (Avg: 92) | Type: Tech

Every so-called “AI memory” product I’ve tried feels like a thin wrapper on RAG.

Engramme doesn’t.

They’ve built a new architecture from the ground up:

Large Memory Models Proactive recall Memories anchored to real events Near-zero hallucinations Works across Gmail, Zoom, Slack, even Meta glasses This isn’t just another feature.

It feels like a whole new category.

Engramme’s Large Memory Models (LMMs) empower every app with persistent memory.

Google solved search. OpenAI solved language. Engramme solved memory.

Join beta: https://t.co/iN6ZgEVSxH https://t.co/LqPxweldnh

See 5 related tweets

@omarsar0: Advanced memory systems unlock agents unlike anything we've ever seen.

This Large Memory Model (LM...

@Scobleizer: Next year Apple (and everyone else) is bringing glasses and pendants to us.

When I talk with "normi...

@svpino: A Large Memory Model (LMM) is a completely new architecture.

An LLM compresses the world's text int...

@Parul_Gautam7: From retrieval → to recall.

Engramme is trying to rebuild AI around memory itself, not just respons...

@TheoBuildsAI: RT @EngrammeHQ: Persistent memory is the Achilles heel of AI.

Engramme’s Large Memory Models (LMMs)...

17. StockSavvyShay (Group Score: 132.9 | Individual: 32.2)

Cluster: 6 tweets | Engagement: 2204 (Avg: 495) | Type: Tech

OpenAI missed a target and the market immediately started pricing AI like nobody else needs compute. https://t.co/aUQbrcJqAf

See 5 related tweets

@StockSavvyShay: This OpenAI headline feels more like a company-specific expectations reset than a systemic warning t...
@StockSavvyShay: OpenAI missed internal revenue and user targets (including its goal of reaching 1B weekly ChatGPT us...
@StockSavvyShay: AI hardware stocks are selling off on OpenAI missing internal targets but $GOOGL Cloud’s CEO just sa...
@BusinessInsider: A report that OpenAI missed on internal growth targets raised fresh concerns about AI demand. Here's...
@axios: AI investors stay bullish after OpenAI revenue miss https://t.co/3jpRrQNl1j...

18. HarryStebbings (Group Score: 132.7 | Individual: 34.2)

Cluster: 4 tweets | Engagement: 108 (Avg: 192) | Type: Tech

Why investors need to give CEOs better comp packages:

"We went public in 21. First year, the stock went up to about 40BN market cap. In 2022, we fell 92% to a little under $4 billion market cap.

For the life of the company, I had only taken equity. I had taken no compensation. At the bottom in 22, I made a decision and that was for the first time to ask for compensation.

I felt like I'm public. Turning this company around is a big task, and I'd like to align myself with investors to say, I'm gonna get paid, but I'm only gonna get paid if the stock recovers.

In order for me to get paid anything, the stock had to clear a certain price.

CEOs who are founders originally started a business taking really big risk. If the belief is that the CEO should then never get compensation ever again, it's completely flawed logic." Adam Foroughi, @AppLovin

What are your single biggest lessons/advice on this @vkhosla @bhalligan @infoarbitrage @taavet @elonmusk?\n\nQT @HarryStebbings: Adam Foroughi is by far one of the best CEOs I have ever interviewed.

Candidly he is insanely stern and cold but also obsessive, focused and utterly brilliant.

With Adam there is zero fluff, like none. He says what he means and means what he says.

Applovin does $10M EBITDA per head.

They have 80%+ margins.

They do $5.48BN in revenue.

No business on the planet has numbers like Applovin.

Following the discussion, I wrote up my biggest lessons from sitting down with him and summarised them below:

Are People Ready for the AI Future That Is Within Every Company?

True AI integration requires a massive "leveling up" of talent. Companies must be honest about the path forward: keeping employees who fail to adopt AI creates a "blockade" to reaching a truly AI-native state. Consequently, we should expect continued tech layoffs as organizations prioritize efficiency over legacy headcount.

Biggest Advice on Token Budgeting and Token Maxing?

Treating tokens as a simple budget or leaderboard is "flawed logic". If you incentivize raw usage, teams will simply create high-volume "crap" that burns capital without driving revenue. Instead, optimize for specific KPIs where token consumption aligns directly with business growth; when revenue is on the other side, the "budget" mindset disappears.

Can You Have a Team Full of Only A Players?

An organization cannot thrive if A players are surrounded by B, C, or D players. AppLovin slimmed its HR department from 80 people to 15 by retaining only "doers" who don't get bogged down in the process. The goal is a lean culture of individual contributors who want to make a difference without needing heavy management layers.

Do the Majority of Company Teams Need to Be Rebuilt for the Technology We Have Today?

If a role is likely to be automated, or if a department is too slow to adopt AI, it is time to rebuild that organization from the ground up. Foroughi cut staff by 40-50% in most departments during a year of triple-digit growth to force the organization into an automated, efficient state.

Why Investors Need to Give Ceos Better Comp Packages

Founders take massive risks to build something out of nothing, and they need continued upside to stay mentally motivated. If a CEO is expected to work without performance-based incentives, they may drift toward new ventures rather than staying committed to the "lonely, stressful" task of scaling a public company.

Why This $160 Billion Company Does Not Have Any Learning and Development

Structured L&D is often disconnected from the reality of high-performance work. The best employees are curious enough to figure things out on their own. By documenting all communication in transcripts and chats, new hires can use AI models to summarize tribal knowledge and develop themselves more effectively than any formal training program.

(links below)

See 3 related tweets

@HarryStebbings: What No One Knows That Everyone Should Know About Being CEO of a Public Company:

"The other thing t...

@HarryStebbings: The Best Founders are Not Motivated by Money:

"Founders don't tend to be motivated by money as well...

@HarryStebbings: Do the majority of company teams need to be rebuilt for the technology we have today?

"Yes, teams n...

19. TFTC21 (Group Score: 132.5 | Individual: 31.9)

Cluster: 6 tweets | Engagement: 4 (Avg: 75) | Type: Tech

OpenAI missed its internal revenue and user growth targets for Q1 2026. Wall Street is panicking. "AI bubble" headlines are everywhere.

Read the fine print. The demand didn't disappear. It moved.

Anthropic's annualized revenue hit $30 billion in April, up from$ 9 billion at the end of 2025 and $1 billion at the end of 2024. Claude now holds 29% of the enterprise AI assistant market, up from 18% in 2024. 70% of Fortune 100 companies use Claude. Over 500 businesses are paying Anthropic more than$ 1 million a year, up from roughly a dozen two years ago. Claude Code alone is generating $2.5 billion in annualized revenue.

The WSJ report itself says it plainly: OpenAI "lost ground to Anthropic in coding and enterprise markets."

This isn't demand for AI cooling off. This is demand shifting. Anthropic is eating OpenAI's lunch in the segments that generate the most revenue per customer. Enterprise clients are moving. Developers are moving. The money isn't leaving AI. It's leaving OpenAI.

The AI bubble narrative assumes the entire market is overbuilt. The actual data says the market is growing fast and one company is taking share from another. That's not a bubble. That's competition working exactly as it should.

See 5 related tweets

@zerohedge: "OpenAI Chief Financial Officer Sarah Friar has told other company leaders that she is worried the c...
@Hesamation: the AI bubble is far more brittle than you realize, and its pop is dangerously entangled to OpenAI....
@zerohedge: "Equities are being driven by one thing…AI spend." - Goldman Delta One

oops\n\nQT @zerohedge: "Open...

@TheRealAdamG: RT @DivesTech: OpenAI has been tracking very high demand on both the consumer and enterprise front a...
@TheRealAdamG: RT @firstadopter: OpenAI says: "The business is firing on all cylinders"

Exponential surge in AI co...

20. Parul_Gautam7 (Group Score: 132.1 | Individual: 36.0)

Cluster: 4 tweets | Engagement: 106 (Avg: 98) | Type: Tech

This guy automated his sales with an agent.

Prospecting, outreach, content, follow-ups, all handled by AI.

The next unicorn might have more agents than employees. https://t.co/gbp7mrSjZ7\n\nQT @emmettmiller_: i built an AI that runs my go-to-market. writes blogs, finds leads, emails users

its basically a junior marketing hire that runs 24/7

saved me 20+hrs/week 10x cheaper than hiring a GTM human

live now @ miniloop [dot] ai

rt & comment "miniloop" for 2m credits on me https://t.co/BjwxnUT3x4

See 3 related tweets

@heynavtoor: every founder I know spends 60% of their time on GTM busywork they hate.

prospecting. follow-ups. l...

@shiri_shh: “POV: You just hired an AI junior marketer that works 24/7

It finds leads on Apollo, writes perso...

@ycombinator: RT @emmettmiller_: i built an AI that runs my go-to-market. writes blogs, finds leads, emails users ...