This is INSANE Democrats NY have passed a bill changing terms such as "mother" and "father" to "gestating parent" and "non-gestating parent." It will now head to Gov Kathy Hochul to be signed into law. Democrats are destroying what it means to be a mother and father.

Everyone is debating Google vs Nvidia. The smarter question is who supplies both of them. Networking. Memory. Optics. Power. These are the companies that win regardless of which chip architecture dominates inference. The platform war is great for the picks and shovels trade.

How did Ferrari build the first sports car that looks AI generated? Looks like a Duracell brick and a camping lantern had a child. Absolute identity crisis

Every infrastructure revolution has victims. 800VDC quietly kills an entire layer of datacenter equipment: → AC floor PDUs: eliminated → Low-voltage UPS systems: progressively obsolete →480V switchgear between transformers and PDUs: gone → Traditional LV transformers: replaced by solid-state Total electrical cost per MW barely moves ($3.6–4.8M band). But where that money flows changes completely. Grey space shrinks. White space wins. Incumbents who don't adapt don't survive. The suppliers who understand this transition before the market does are going to look very smart in 3 years. $GEV

Bloom Energy went from “clean energy stock” to core AI infrastructure play. The market finally realized: No power = no AI deployment. Hyperscalers can get GPUs now. What they can’t get fast enough is electricity. That changes the entire AI investment landscape. $BE $NVDA $VRT $GEV

For years the AI debate was: bigger model or faster model? Turns out developers will pay a massive premium for faster — even when a smarter model is available. Tokens per second is the new MHz wars. And this time the speed premium prints real revenue. Nvidia just made sure it owns that trade too.

Everyone keeps talking about bigger AI models.But the real bottleneck may now be: memory movement latency power.Cerebras’ wafer-scale design keeps far more data directly on-chip instead of constantly moving it across networking fabrics.The result: extremely fast inference.The AI stack is evolving from: raw compute → efficient token generation.That changes everything:networking memory cooling inference chips power infrastructure custom silicon The next AI cycle may look very different from the last one.

$AAOI earnings in one line: Q1 was fine. 2H is where it gets interesting. Management just guided to 141% full-year revenue growth and raised guidance to $1.1B . 400G products? Up 10x year-over-year. The stock dipped after hours because investors want proof. Fair enough. But if execution matches the guide — this is one to watch closely.

Most people don't realize ARM's IP is already inside almost every AI chip being built. Graviton. Axion. Vera. All ARM architecture. Now they're stepping out from behind the curtain and building their own chip to compete directly. CEO thinks ARM will be the #1 CPU architecture by end of the decade. The long term story here is one of the most interesting in semiconductors right now. $ARM

$COHR just reported: • AI data center revenue 41% YoY • Earnings 55% YoY • Orders coming in 4x faster than they can ship • Nvidia invested $2B into the company The stock is still under the radar. The AI infrastructure buildout isn't just GPUs.

$LITE by the numbers this quarter: → Revenue: $808M ( 90% YoY) → Next quarter guide: $1B ( 105% YoY) → Gross margins: 13 points year over year → EPS: 316% YoY → Supply-demand gap: >30% and widening They are selling everything they can make and still can't keep up with demand. $2B/quarter is the target. They think they get there in under 2 years.

Astera Labs was supposed to grow 8% last quarter. They grew 14%. Revenue nearly doubled year over year. Margins beat guidance. Free cash flow up 5x from last year. The market keeps underestimating this one.

AMD Q1 by the numbers: ↑ 38% revenue ($10.3B) ↑ 43% earnings ↑ 57% data center ↑ 3x free cash flow — record $2.6B Server CPU: 4th consecutive record quarter. Q2 guide: server CPUs growing 70% year over year. The fundamentals are not the problem.

Reddit Q1 by the numbers: 📈 Revenue 69% YoY 💰 Free cash flow up 2.5x YoY 🏦 $2.8B cash, zero debt 📊 Rule of 40 score: 109% Wall Street is worried about growth slowing to 44% next quarter. Meanwhile the fundamentals just keep stacking. This one's undervalued.

Google cut their own AI serving costs by 78% in 2025. Now they're passing those savings to customers while expanding margins. Cheaper prices → more customers → more scale → lower costs. The flywheel is spinning. $GOOGL

Everyone's focused on Oracle and Bloom Energy. But the real story: hyperscalers and cloud providers now make up more than half of Bloom's AI data center pipeline. Oracle was just the beginning. $BE

The AI buildout has a power problem. Data centers need electricity the US grid wasn't designed to deliver. GEV makes the turbines, transformers, switchgear, and substations that fix that. Q1: profits doubled. Orders surging. Backlog $49B. This is early innings.

🚨 JUST IN: Rep. Chip Roy is filing the "MAMDANI Act" to DEPORT and DENATURALIZE foreigners advocating for or linked to an Islamist, socialist, communist or Chinese communist party YES! We need this! The bill also targets aliens belonging to organizations advocating for Islamic fundamentalism, Marxism, socialism or communism, per Breitbart On top of deporting and denaturalizing, the legislation BLOCKS admission to this category of aliens as well 🔥 I support this 100%. MAKE IT LAW 🇺🇸

$DELL generated $5.1B in free cash flow last quarter. Same quarter a year ago? $474M. That's not growth. That's a different company. AI server revenue up 342% YoY and margins held near 3-year highs despite the headwinds. Storage attach rates are the secret weapon keeping this thing profitable.

Nvidia’s next leg higher won’t come from bigger models. It comes from making AI usable at global scale. Rubin isn’t just a faster chip — it’s built for: • Memory-heavy workloads • Massive bandwidth • Distributed inference The bottleneck is shifting: Compute → Memory → Networking → Power And Nvidia is positioning to control the entire stack. When inference scales, it pulls: AVGO, MU, COHR, NET, GEV… the whole ecosystem. This is why IOF calls it: 🧠 “The Inference Kingdom Expands” $NVDA