Gromker

trader | redart | Journal

[skinky (afk) @goatdsalmon]

Co-founder of Jito btw

Don't let one of the most obvious Solana plays slip right under your nose.

[Chapo @ChapoGrimey]

if you must, go poor buying this stock, by all means im not giving you financial advise. i'm only showcasing my bids.

$ampg - anything under $10. is free.

[washed @notwashed]

Allowing yourself to hate crypto more cus of onchain shenanigans, at this stage of the bear and at these prices, is pretty stupid imo

Eerily reminiscent of the celeb meta in summer of 2024. Lots of ppl got chudded out cus of this, then just a few months later we had one of the greatest wealth creation events in the AI meta

We’ve had approximately 9 months to hate coins, starting to hate them more cus of this airdrop meta is short sighted. Price going up and greed can change things very quickly

[Chapo @ChapoGrimey]

The transition from AI to robotics is hitting a major bottleneck and the market still hasn’t fully priced in that robotics will require enormous data center capacity to train, operate, and scale effectively. While everyone’s focused on AI compute demand, the next wave of physical intelligence is going to be just as data hungry.

I’m aggressively bidding data centers. Nothing can convince me otherwise.

$GLXY is positioning itself exceptionally well. They’re set to build one of the largest data centers in the world, and they now have two operational facilities nearing full tenant announcements. One of these sites is located directly adjacent to a new plot of land right next to $SPCX ’s launch center and headquarters.I strongly suspect SpaceX will eventually partner with or utilize Galaxy’s data centers. The geographic proximity and direct line-of sight communication potential make it a highly logical fit.

[asparagoid 🐂🀄️ @asparagoid]

Similarly when I lived as a Buddhist monk for a few years there was a phenomenon I jokingly referred to as Virgin Monk Syndrome

When I could tell someone was there (long retreats, etc) out of resignation and lack of life experience, but wasn't actually bored of materialistic desire, it was more that they'd given up or were too afraid of rejection

Same guys always sat on the cushion having endless sexual fantasies, with materialistic visions in the back of their minds, etc. When you talked to them, you could tell they hadn't lived much yet. So things still held appeal, because they hadn't yet satisfied them, and this made their unconscious mind still drunk on the idea of it

So they were there for long retreats, but at the same time, not really, not fully. I would say many were just afraid of women, or afraid of failure in business, and so on. It's not that those things are pointless, it was that you can't give yourself to a path (like hardcore insight meditation) path fully before you take the other ones to their end. Because strength of intention is everything. Because something in you will always be saying "what if?" and something in you will know, deep down, that you are really there out of avoidance, or to fill time, or because you are lost

Better to start where you are, get rejected a hundred times, get the girl, maybe even get the yacht, then see what's left when you look inside your soul

It's like God gives you a fear, and he's going to make you face it no matter what. You can't run away. If you sit on the cushion to avoid the fear, he'll distract you with a million lustful fantasies, because it's not actually the most important thing for you to do for your spiritual development

Such guys also ended up quitting earliest

Virgin monk syndrome

[Electra @Electrarythm]

Most people never realize the most powerful moments for changing their mind happen twice a day, right as you fall asleep and right as you wake up.

Right at those moments your brain is in a slow, drowsy brainwave state, the same one you pass through during light sleep.

The part of your mind that normally judges and rejects new information hasn't fully come online, so when you picture something then, it reaches your subconscious directly instead of getting blocked.

That's why the same visualization works far less in the middle of the day. By then your analytical mind is fully active again, so it's quicker to reject what you picture.

So at night, do it as you're falling asleep. In the morning, do it the second you wake, before you reach for your phone or get out of bed. Keep your eyes closed and go straight into it.

Pick one specific scene, not a vague goal. Don't try to picture "success." Pick one exact moment that would only be real if you already had what you want, sitting in the driver's seat of the car you want with your hands on the wheel, checking your bank account and seeing the number you've been working toward, or walking into your new place for the first time with the keys in your hand.

One moment you can step into.

Then build it one sense at a time.

Sight. Look at the scene through your own eyes, not from the outside. What's in front of you, the colors, the light, the people, the details around you.

Sound. Add what you'd hear. The voices, what people are saying to you, the tone of their voice, the noise of the room, music, traffic, whatever fits.

Touch. Feel the physical things. The seat under you, the wheel in your hands, the floor under your feet, the temperature of the air, your clothes against your skin.

This is the part most people skip, and it does the most to make your mind accept the scene as real.

Smell and taste. Add them if they fit. The smell of the new car, coffee, food in front of you. These reach the most emotional part of the brain, so even a faint one pulls you deeper in.

Then add the emotion. Once the scene is built, feel what you'd actually feel if it were real, the gratitude, the confidence, the pride of it being done.

This is the most important part. Without the emotion it's just a daydream. With it, your brain stores it closer to a real memory.

Hold that feeling and let it sit in your body.

Then stay in it. Don't describe it in words or repeat affirmations. Just be in the moment and let it run for a few minutes.

Keep the scene short and do it every day, because what makes it work is doing it consistently, not how long you sit there each time.

Always do it through your own eyes, never watching yourself from outside. And finish on the emotion, since that's what stays with you after you get up.

[Peter @Onepeterrr]

“• buy coins that have established communities after the bundlers are already out

the thing with launches now is that every coin is bundled, but ppl bundling coins have very short time horizons, they aren't going to wait a couple weeks post launch to sell, so there's a window where protocols & memes have already established themselves and theres much less risk “.

He said it all here. These bundlers can’t even hold for long. Once they’re out, that’s when you buy. They’ll just move on and deploy another one.

If we can all be a little more patient like we used to be, things will start changing again.

[Ansem 🐂🀄️ @blknoiz06]

this is the first cycle solana infra plays will rip alongside sol, first cycle was too new, second cycle unlocks & everything, this time have good core protocols that are essential to the ecosystem & will benefit from increased retail + institutional activity on chain

[Ansem 🐂🀄️ @blknoiz06]

Everybody's gonna call me insane BUT solana:So11111111111111111111111111111111111111112 from $60 to $600 this cycle

Last cycle it hit $297 on memes alone. that was the proof of concept

This time solana has memes, breakout consumer apps like CARDS, AND a real perps competitor to hyperliquid

https://t.co/G5ZPk8jTOi

[washed @notwashed]

trading attention is tough these days. you think you found something new, only to later realize it exists entirely within your little echo chamber

very rarely do things actually go viral or catch steam like they used to - engagement is misleading these days. you see something with millions of likes, but when you ask people if they've heard of it they look at you like you're crazy. trends and memes still go "viral," but they're often only viral within your hyperspecific algo

fyp algos are more fragmented than ever. 5 years ago it felt like there was one big internet everyone shared. now it feels like millions of distinct feeds occasionally overlap. and when something overlaps? cool, it lasts 48 hours before everyone moves on to the next thing

monoculture is dying in purely online things, but getting massive in unreplicable experiences like sports and movies - less personalized areas

there's definitely a trade here, just dont what yet

[WR☻NGUSER ✗ @wronguser000]

If i had sub 500K port i'd be scalping the main onchain coin all day all night, rn its solana:CARDSccUMFKoPRZxt5vt3ksUbxEFEcnZ3H2pd3dKxYjp tomorrow could be anything

Simply stay glued to your screen with Bitcoin on one half and Cards on the other and just slam the buy button whenever BTC gets some peace

Scalp for 10-20% rinse and repeat

Thats literally how i made most of my money trading memes from the WIF POPCAT era

Consistency, patience and never accepting losses will lead you to compound your wins on top of eachother and one day you wake up and you 10xed your trading pf while never hitting any "big" runners

Interest will translate into volume and volatility, cherish it and extract as much as you can

[Electra @Electrarythm]

Once you become the money guy, your brain starts filtering the entire world through it. The deals, the opportunities, the people who can get you ahead, that's all you really see. Everything else barely registers, because your attention is locked on one station. The options that don't fit the label could be sitting right in front of you and you'd walk straight past them.

And the second something threatens that label, it feels like a threat to your survival. If you're the smart one, you can't touch anything you might fail at, because failing wouldn't just be failing, it'd be proof you're not who you say you are. If discipline is your identity, you can't rest without guilt, so you grind until you burnout. The label stops being who you are and becomes the cell you're locked in.

It also forces you to bury half of yourself. The spiritual one isn't allowed to be angry or want money, so he buries it. The tough guy isn't allowed to be sensitive, scared, or open, so he hides it from everyone, because that's "not who he is." So you live as a fraction of yourself, everything that doesn't fit shoved into the dark, and you call that being you.

And when the role finally goes, it doesn't feel like losing a job or a hobby. It feels like dying. The businessman who loses his business doesn't just lose money, he loses himself, and that's why so many of them end up at the bottom of a bottle when it falls apart. There was no "him" left underneath it. His whole existence walked out the door with the thing.

It even picks your life for you. The same kind of work, the same kind of people, the same kind of days on repeat, and you think you're choosing it. You're not. The label is.

The way out is to stop being what you do. You can do business without being a businessman. You can be spiritual without becoming "a spiritual person." You can love the game without turning into the sports guy. You can do the thing without ever becoming it.

[Ansem 🐂🀄️ @blknoiz06]

@Flaming0x i believe kintara will be one of the apps on sol that lead interest coming back into crypto alongside cards, pumpcade, some ai protocols, & whichever memecoin gains the most momentum in Q3 above 100M mcap

[DarylTanky @DRTnky]

I’m a fan of Felix now

He touches on a lot of points that I’ve mentioned before re Singapore, namely:

- We are paid to optimise, not innovate
- We celebrate wins, but sh*t on losses
- Singapore is a great country to run operations, not R&D labs

The outcome of that is a service focused nation that is probably great at driving things from 90 to 100

Not 0 to 1

Not 1 to 10

Not even 10 to 50

We only do great when there’s already a playbook

Now, granted, we are a small nation and every mistake could be alot more costly

I do wish we get to see more innovation but truth of the matter is that capital already realised that Singapore isn’t the place for start ups

Start up VCs are already exiting APAC with funds like Granite not even deploying much in the region

Private Equity are only focused on late stage infrastructure businesses now - they know money in Singapore isn’t to be made not in the innovation, but operations

I do hope that we get to see more risk taking, more innovation, more fun

I love Singapore but objectively, we are not even close to America when it comes to reinventing the wheel

[based16z @based16z]

https://t.co/LI8UiKx84o

[xet @xet]

i have lost it all ( millions of dollars ) and im quitting crypto forever. all i ask for is 2 minutes of your time before you lose it all as well.

i joined crypto during the first bear market many years ago, when bitcoin crashed 85% and many alts disappeared forever. back then crypto seemed like something “mysterious”. second bear market, when ftx / luna collapsed, was also COMPLETELY different from what is happening now. im not even talking about the price action. we are not even in a bear market yet btw, btc will be at 28k soon.

crypto’s image and reputation are destroyed. this is not what the last 2 bears were like. everything has changed. liquid money is split between to many different sectors.

let’s start with this, you cant even scroll any social media platform without seeing a gambling ad or prediction platform. these influencers always find an excuse to post that promotion and get away with it.

“i lost 500k trading, now im gambling to make it back. i made it.”

translating: no he didn’t make SHIT. he has no skills. he got paid for the promotion and his session luckily hit. they want to get you on a hook and lose every single dollar you own. these guys genuinely have NO money. NO MONEY.

degenerate liquidity is split between 10 prediction platforms, 1000 gambling websites, and 10,000,000 memecoins. there is NO liquidity left to satisfy everyone. it will ONLY get worse. during the last bear, it was 1 platform, 100 websites and 10k memecoins. even if liquidity REMAINED the same, it has now split another 10x.

crypto and memecoins have a terrible reputation. people DO NOT trust this space anymore. even i don’t trust solana tokens anymore. i have made tens of 500x and 1000x calls, but what i see now is just taking a piss, coming back 5 minutes later and seeing a token at -80% in SECONDS.

you really think these “rich traders” have money and post gambling bullshit for fun? no they don’t. and im here proudly saying that I HAVE LOST IT ALL AND IM NOT HERE TO LOSE MY INTEGRITY BY PROMOTING A LIFE KILLER. and im saying this for everyone. just admit you are broke, too hard? you know everything yourself bro. stop pretending ur a millionaire.

to solana traders, you guys really think there will be another run? who is coming to bid your tokens? are you absolutely blind?

ansem posting pumpfun fees and low mc tokens again is the ultimate top signal for solana. there is no one bigger than him in terms of attention and reach. i like his approach though, he is always pushing real belief, but bundlers won’t ever listen as they are in a comfort zone and real “community members” are out of money cause bundlers took everything. it will only get worse. each shill will get less and less eyes. how exactly does this get better?

every token has hidden bundles that nuke 24/7. there are MILLIONS of old tokens. there is simply NO liquidity to make everyone happy. quit before it’s too late. RUN.

was fun. this is it for me.

trading nowadays is self torture. legit developers get destroyed because there is no buy pressure after the first 8 hours of a project, while bundlers keep winning because they control the narratives. real builders get farmed and forgotten.

the game is finished.

[washed @notwashed]

im broken record atp, but ik im right, so ill keep yapping about this

the meme trade is no different from the shoe reselling craze from 2014-2021 - seeing more and more similarities every day. ofc nike is bigger than just shoes, but shoe hype massively helped the brand win. nike did something new when they collabbed with kanye and virgil, but never did anything new after those collabs. after failing to innovate, they just milked the big names dry and released the same silhouettes over and over again until they became stale. it was normal for hyped releases to resell for over 1k, but as releases grew more stale, no one wanted to pay the same premium anymore. the same thing is happening with memes. 100m coins were once normal, but now 10m is the new 100m. nothing new or exciting has launched lately, it's all recycled themes and narratives

with that said, i still think memes are sort of evergreen. there will continue to be viral moments and new things to speculate on, but the ceilings are just lowered. bringing back believing, raiding, cults, etc., doesn't change anything. i think the sol fnd has noticed this, hence they don't show memes a ton of love anymore. they want to be the onchain nasdaq - it doesnt seem like they have much interest in being known as a vaporware factory anymore. personally, im following their lead and not trying to fight the trend

[Ansem 🐂🀄️ @blknoiz06]

this happens at the beginning of every cycle

it's not that im trying to onboard them through memes its moreso thats just what i know retail is going to trade onchain, sure there are also great protocols but most of them are facilitating trading & speculation

solana is the only coin in crypto today that benefits from degenerate gambling on memecoins and also degenerate gambling on stocks and RWAs

if you have $100 extra dollars in your bank account are you going to buy apple stock or a dogcoin?

[Ansem 🐂🀄️ @blknoiz06]

i know theres more home run trades in the ai open source models breakout trade & also in the memory optimization trade

so far ive been shilled

• minimax hk:0100
• zhipu: hkg:2513
• penguin solutions: $PENG

[Saso Capital @saso_capital]

Context windows are the most underappreciated demand driver in the AI trade.

Two years ago, GPT-4 shipped with 8K tokens. Today, Gemini runs at 2M. Claude at 1M. GPT-5 at 1M+. That’s a 125-250x expansion in the amount of active memory a single model session consumes, and we’re still early.

Micron’s EVP Manish Bhatia told JPMorgan on May 20 that context windows are growing at roughly 30x per year. Every token in a context window generates key-value (KV) cache data that has to live somewhere in memory during inference. Longer context means the KV cache balloons. A 1M-token session on a large model can consume hundreds of gigabytes of KV cache alone, per user, per session. Multiply that by thousands of concurrent users running agentic workflows where models call other models in chains, and the memory required per GPU cluster doesn’t scale linearly. It scales combinatorially.

This is why inference is 70% memory-bound and only 30% compute-bound. The GPU isn’t the bottleneck anymore. Feeding it is.

Two stocks sit directly in this path.

$MU Micron makes the DRAM and HBM that stores the KV cache at the hardware level. Every GPU needs more memory stacked on top of it with each generation, and it’s still not enough. HBM4 is ramping 2x faster than HBM3E. Fiscal Q2 revenue was $23.86B (+196% YoY). Guided ~81% gross margins. Entire 2026 HBM allocation sold out under fixed-price contracts. First 5-year strategic customer agreement signed. Five fabs under construction. Customers are receiving only 50-67% of the memory they’re requesting, and Bhatia says supply won’t catch up for the foreseeable future. Forward P/E under 8 for a company printing 196% revenue growth. 18% from a trillion-dollar market cap.

But HBM capacity on-chip tops out. A single GPU can only hold so much. That’s where the architecture has to extend outward, and that’s where PENG lives.

$PENG Penguin Solutions launched the industry’s first production-ready CXL-based KV Cache Server in March. It’s an 11TB memory appliance purpose-built for exactly the problem context window growth creates: KV cache data that overflows the GPU’s onboard memory. Instead of recomputing that data from scratch (expensive, slow, wastes GPU cycles), Penguin’s MemoryAI server stores it in a disaggregated CXL memory pool with 10x faster access than NVMe. Compatible with NVIDIA’s Dynamo framework out of the box.

This is the missing piece in the inference stack. When context windows were 8K, KV cache fit inside the GPU. At 1M tokens, it doesn’t. Someone has to architect, build, and manage the external memory layer that catches the overflow. That’s PENG’s business, and nobody else offers a production-ready CXL KV cache product at this scale.

The numbers back it up. PENG beat Q2 expectations and raised full-year revenue growth guidance to 12%. Deployed 99,000+ GPUs across customer sites including SK Telecom, Shell, Sandia National Labs. Stock at all-time highs, up 172% YTD, $2.7B market cap.

Here’s how the flywheel works:

Longer context windows → larger KV caches → more DRAM/HBM consumed per session → $MU sells more chips at higher prices.

KV caches overflow GPU memory → need external disaggregated memory infrastructure → $PENG deploys and manages that layer.

Agentic AI compounds both: agents run multiple model calls in chains, each maintaining its own context, each generating its own KV cache. One user session can spawn dozens of parallel inference threads. The memory footprint per “task” isn’t one model’s context window, it’s many, running simultaneously.

Context windows are only going in one direction. Models don’t get shorter memories. Users don’t want less context. Agents don’t run fewer steps. Every advance in AI capability translates directly into more memory consumed per unit of compute, and that’s a structural tailwind for both the chip maker and the infrastructure company that turns constrained memory into deployed, managed capacity.

I’m concentrated in $PENG and $MU.