Omri Barzilay

Morgan Stanley estimates the humanoid robot market could reach $5T by 2050. Bookmark this. Robots don’t scale without this stack: • Bearings / Actuators → $RBC $AME $RRX $NOVT • Sensors → $CGNX $ALGM $STM $OUST $VPG $MBLY $AEVA • Edge AI → $AMBA $CEVA $LSCC • Humanoid Semis → $NVDA $AMD $QCOM $TXN $ON • Motion Control → $ROK $PH $ALNT $ABB • Power Semis → $NVTS $VSH $MPWR • Connectivity → $CSCO $IRDM $SLAB • Rare Earths (Motors) → $MP $UUUU $USAR • Key Suppliers → 002050.SZ (Sanhua), 688017.SS (Leader Drive) Own the bottleneck. Follow the right people on X.

@Jason I’m in

We started an AI founder twitter group... reply with "I'm in" if you're a founder and want to be added

@NirHirshman מדהים! כמה טוב העולם הולך להרוויח

בום 💥 ראש ה -SEC מטיל פצצה: הויכוח שקרע את עולם הקריפטו מסתיים היום! בהודעה תקדימית, ה-SEC מפרסם היום כללים ברורים וקובע כי אלה **לא נחשבים ניירות ערך**: 🔹 סחורות דיגיטליות (ביטקוין, אית'ר ועוד מגוון מטבעות) 🔹 פריטי אספנות (NFT) 🔹 כלים דיגיטליים (כרייה, סטייקינג, דיפיי) 🔹 סטייבלקוינז (שהוגדרו תחת GENIUS) למה זה חשוב? במשך שנים כל מי שהנפיק מטבע דיגיטלי, או פיתח כלים בתחון מצא את עצמו תחת איום של תביעות ורדיפה ע"י רשויות שטענו כי הוא מפר את דיני ני"ע. יזמים וחברות נאלצו לברוח מארה"ב (וגם מישראל) להקים מבנים מסובכים במדינות עלומות, וכל זה כדי שיוכלו לפתח את טכנולוגיית המחר. נגמר. בהחלטה אמיצה ה-SEC מפזר את הערפל, ונותן ודאות לשוק, כדי לקרוא ליזמים- תבואו לפתח באמריקה. אנחנו חייבים ללמוד מהם. ישראל חייבת לקרוא ליזמי הקריפטו לחזור הביתה, ועכשיו.

@LucidCap @AdamRFisher @yanaioron @liadagmon הסרט מציג כאילו שהמצב הנוכחי של הפיתוח ב2030 ישאר בערך אותו דבר כשאנחנו צפויים לקפוץ ביכולות 10X לפחות כל שנה

מניות ה SaaS עולות בלהבות, $MNDY נסחרת בשווי הקפיטריה של העובדים, וול סטריט לא מאמינה ש $WIX תהיה פה עוד 6 שנים. יצאנו למסע בזמן עם אדם פישר @AdamRFisher , ינאי אורון @yanaioron ליעד אגמון @liadagmon ויאיר שניר (DTC) ביחד ניסינו לצייר את עולם התוכנה ב 2030: 00:00 רקע ופתיחה 02:29 האם יש באמת יתרון ל AI-Native? 07:55 האם חב' יבנו בעצמן SaaS במקום לקנות? 15:34 אינפלציה של חב' תוכנה? 21:15 האם UI עדיין רלוונטי בעידן האג'נטים? 23:11 מה יקרה למספר המפתחים ואיך האקו-סיסטם בישראל יושפע?

@levie Here is a good example of this motion

There’s a window right now where AI agents will get built for every vertical and domain. The playbook is to go deep on the context engineering required for the vertical or particular space, figure out the right UX that ties into the existing workflows naturally, and connect to the relevant data sources and tools. Especially early on, it’s useful to get as close to key customers as possible to figure out what’s working and what’s not and constantly make improvements to bring them back to the mothership. AI is moving so fast right now that there’s a huge premium in making quick updates and seeing how they improve the customer’s workflows. It’s also important to price the agents for maximum adoption with simple subscription prices or on clear consumption model, and expect to ride out the cost improvements from AI efficiency. Don’t get too greedy on price right now as market share is likely most important. It can be helpful to go after use cases that are constrained by the availability or high cost of talent. This means that any incremental boost in productivity in these spaces offers high ROI for the customer. In these areas, customers will always be willing to try AI agents to finally get around to solving their problems. This is why AI coding agents, security agents, or legal agents are taking off right now initially. These are all areas where demand for solving the problem has always exceeded the level of talent available. But every vertical has examples of this. There’s a clear moment right now where the next generation of these AI Agents will get built across every space.

@haridigresses @bgurley Direct listings = “democracy” 😂 2020-21 score: WRBY -56 % (54 → 24), ASAN -46 % (27 → 15), COIN ±0 % (381 → 378), RBLX +114 % (65 → 138). Basket +3 % vs S&P 500 +64 % (3,875 → 6,339). Same Wall St. shell game—retail = exit liquidity. Enjoy the single-share crumbs 🎉

Time for a quick rant on the deeply broken IPO process. @bgurley has been beating this drum for years, but I don't think anything paints the picture quite as clearly as the Figma IPO. Figma tripled in its IPO debut, leaving ~$2B+ on the table. That value solely accrued to the investment bankers (who provide "price support" during the listing, what a joke), and institutional investors (yes some of this is Blackrock and your retirement ETF, but let's be real most of it is not). Retail investors get left in the cold... Robinhood investors are posting that they got one share, instead of the 600 or 1,000 shares they tried to get. Even HNW investors (who are not UHNW) got zilch... apparently the vast majority of Morgan Stanley PWM clients got no allocation either (and they were the lead underwriter!). This IPO was so obviously mispriced that it should be considered the equivalent of investment banking malpractice. Don't even get me started at the charade of "oh we'll start at $25-28; no no it's now $28-30; never mind it's $33". You might say: "but Hari, how could they know? they needed to play it safe to avoid a bad showing on day 1." Do I have a response for you... 1. Anyone who's ever done a DCF, or used Figma, or worked at a startup/ could have told have told you $33 was too cheap. My estimate is that they're worth $50 - 90 / share ($55 being very very conservative and $70-90 is probably the reasonable range... and shocker, that's where it's trading at the open). 2. There is basically no correlation between how a company does in its IPO vs. it's long term success. 3. Figma didn't even really need the money, they've been profitable for a while. 4. We have a mechanism for price discovery! It's called a direct listing! It's called private secondary markets! Spotify, Coinbase and others have already paved this way. Figma, for some baffling reason, chose not to follow this path. Between selling stockholders (24M shares) and Figma's offering (12M shares), they left billions of dollars on the table. In fact, Figma studiously banned all secondary transfers for years (I would know, I've been trying to buy into the company for years, been a customer and massive fan since like 2019). The process is even run by bankers, but they make lower fees so they don't like it, of course. Two very simple takeaways: 1. If you're profitable and don't actually need the cash, do a direct listing. Anything else is not the responsible thing to do as a company. 2. If you're a consumer or prosumer brand, you should really carve out an allocation for your customers and fans. Airbnb did a terrific job with this. Figma should have, too. I'm still a fan, but damn would I have loved to support the company earlier in the journey.

@asmartbear Copy-paste-improve

Copying others is a good way to learn. It’s not you yet, and not “success,” but it’s a natural process. For example, early in blogged I copied others' styles. Really trying on, like clothes. Nothing wrong with play. longform.asmartbear.com/learn-by-copy?…

@nivi Fundraising is the invisible “0.5” slot—do it right and the whole stack gets super-charged: you lure better recruits, fund faster product loops, boost marketing firepower, and give sales real ammo.

What matters at a startup, in order of importance: 1. Recruiting 2. Product 3. Marketing 4. Sales -∞. Planning, Meetings, HR, etc. Do any item on this list well, and the next one is “easy”.

@MeetKevon Grok 4’s the appetizer—wait for GPT-5 later this month

9 months ago I never followed or looked at any AI news, now I’m excited about Grok 4 release. Funny times.

Figma’s IPO filing + Grammarly’s grab of Superhuman prove SaaS isn’t dead, it’s mutating. The next wave is “acquire the feature”: AI-native point tools will get scooped up before markets can price them like platforms—build fast, bundle faster.

Microsoft’s latest layoffs make it clear: it’s not age, it’s adaptability. #AI is erasing entry-level grunt work and any seasoned skill that stops evolving. The only safe job description? “Always learning.” Level up or be leveled

@jmj When code is commodity, conviction is currency—but distribution is still king. Killing 99 % of ideas means nothing if you can’t win the 99 % grind of sales & marketing. Product is table-stakes; pipeline crowns the winner.

In a future where engineering costs trend to zero & you can build anything overnight, focus becomes your competitive advantage. Seeing this firsthand: friends tinkering nonstop but paralyzed by infinite choice. Building is easy—deciding what to build is hard.

Skin-in-the-game is a great alignment signal—but it’s not the whole story. A first-time manager with 1% commit and every reputation chip on the line can out-hustle a 10% commit from a multi-fund franchise that’s already playing portfolio defense. True alpha comes from how you deploy capital—focus, discipline, access—rather than just whose capital it is. GP dollars matter, but GP edge compounds

Countdown — Sam Altman says GPT-5 is tracking for summer ’25 release. Prediction markets put odds near 90 %. Time to get your infra in order. Why it matters — Leaks point to: • 10× parameter jump • 200-300 K context window • Native video + 3-D understanding • Better planning / less hallucination Cost curves — GPT-4o halved price; GPT-5 is rumored to cut another 40 %. Every SaaS feature you sidelined as “too expensive to run” just became viable. Agent-mode by default — Expect multi-step tool calls baked into the API. Opportunity moves from single prompts --> orchestration & UX layers. Vertical copilots — PhD-level reasoning means you can automate edge-case work in law, accounting, bio-R&D. Your moat = proprietary data + compliance workflows. Synthetic everything — Higher-fidelity text, audio, video generation lets you create training data, user simulations, even product demos on demand. Infra gold-rush — 10× params = GPU starvation. Startups that squeeze inference (quantization, caching, distillation) will print money. Trust layers — Bigger models invite bigger audits. Build eval dashboards, red-team services, and usage-based guardrails; enterprises will pay. UI shake-up — With video + longer memory, chat UIs feel dated. Expect timeline-style canvases, voice-first flows, and “in-file” AI that remembers sessions. Action items • Structure data now (RAG will feel free) • Modularize prompts → easy swap-in • Budget for 0.1 ¢ / 1K token dreams Reply with the niche you’re tackling & I’ll DM a GPT-5-ready roadmap.

The silent killer of 90 % of side-projects: feature creep. Met a founder who promised to launch after “one last tweak.” 90 days & 27 new features later: users = 0. Dark-mode Easter eggs don’t pay bills. Ship the 70 % version, let customers write v2. How many features did you build before user #1?

@SimonHoiberg Love n8n, but “slap a Lovable UI on it and sell” skips the hard part: support, reliability, and enterprise-grade integrations. Low-code workflows are table stakes now—Zapier, Make, and Retool already chase the same SMBs. Differentiation ≠ prettier front-end.

ℹ️ In my latest YouTube video, I'm covering 2 other ways you could monetize n8n. Watch the full video here: youtube.com/watch?v=foEW38…

There is an incredible business opportunity using n8n & Lovable - and I don't hear anyone talk about it! n8n is getting serious traction. And because of this, YouTube and Twitter are now FILLED with people sharing really amazing workflows. What is a good workflow? It's a problem being solved. Many people find n8n very difficult. So use Lovable to create a simple UI, wrap a useful workflow, and offer this to less techy users. Here's how you could monetize this 👇

@rrhoover @vedikaja_in Fantastic 6-Q rubric. I’d tack on a 7th: Compounding data flywheel—does every user interaction create proprietary feedback that instantly upgrades the product? If that loop isn’t locked down, the next model drop will flatten your moat.

Every investor is asking: "How is this defensible?" @vedikaja_in created a rubric to stress test a startups AGI-proof moat.

@VinnyLingham True, but only if the playing field’s level. Just ~1 M of NYC’s 4.7 M registered voters cast a ballot, and 1 in 5 New Yorkers couldn’t vote at all thanks to closed primaries. Before we say “we deserve it,” let’s fix access and engagement first.

Democracy means that we get the mayors and politicians we deserve - for better or worse…

@MeghanKReynolds @T_Caldwell @stepstonegroup Who crossed 3X TVPI?

Heard from LPs this week: A data present 🎁 So many convos in ‘24 around fund math. Can a big fund deliver big returns? So, I posed the question to my VC data guru @T_Caldwell from @stepstonegroup How many VC Funds over $2B have delivered over 3x Net TVPI or DPI? The Punchline: Only 3 out of 72 $2bn+ VC Funds have crossed 3x TVPI and only 1 has crossed 3x DPI (1%) (All of those funds are pre-2020, which makes sense these are funds that have had a chance to mature) We then looked at all PE, Growth VC Funds… to apply the same question to big funds in general, not just VC The Punchline: Only 15 out of 648 (2%) of $2bn+ Funds have crossed 3x TVPI and only 5 has crossed 3x DPI (1%) I think a better understanding of this data will reset expectations for VC portfolios. Big funds expected to deliver consistency. Small funds expected to deliver upside.