Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
1 selected
Loading...
0 selected
Loading...
0 selected
0 selected
All post types✓
Portfolio updates✓
Stock lists✓
Research✓
News✓
All Content✓
Source feeds
Buzzberg Top 50
✓
All market capsNo capitalization filter✓
200 B and aboveMega✓
10 B to 200 BLarge✓
2 B to 10 BMid✓
0 to 2 BSmall✓
Custom
Enter market cap range in B USD
All directions✓
▲ Long✓
▼ Short✓
⛔ Avoid✓
✂ Close✓
◦ Others✓
Any score✓
LOW+✓
MED+✓
HIGH✓
12:24
Aug 21
Aug 21
Bankless
1d
MRNA 1ST
LITE
COHR
XBI 1ST
CIBR 1ST
▾
HIGH
Personalized mRNA melanoma vaccine validates Moderna.
Moderna and Merck reported positive Phase 3 melanoma trial results for a personalized mRNA cancer vaccine. AI identified patient-specific tumor mutations to trigger an immune response, making cancer treatment personalized. Moderna stock surged roughly 250-300%, and approvals could come as soon as 2027.
MRNA LONG
AI trade shifts to photonics/optics/power.
Ejaaz sees the next AI investment theme shifting to the photonics, optics, and power stack. NVIDIA is moving that way, having invested heavily in Lumentum and Coherent, and is now making a major play on power-side data center infrastructure.
LITE LONG
COHR LONG
Cheap sequencing plus AI accelerates biotech.
Biology is becoming an AI and software problem. Genome sequencing costs collapsed from billions of dollars and years of work to under $1,000 and hours, AI context windows can hold DNA sequences, and this enables hyper-specific personalized treatments. Josh sees biotech at a major inflection.
XBI LONG
AI threats will drive cybersecurity demand.
AI models consistently beat human-written code and are especially strong in verifiable domains like cyber. The threat range includes AI-written malware, self-replicating prompts, and reasoning worms, so Josh expects a lot more cyber-related activity and demand ahead.
CIBR LONG
Unitree IPO looks overhyped and fragile.
Unitree's IPO was oversubscribed 5,500% and shares rose 542% on day one, but Ejaaz sees overhype because 40-60% of its humanoid robots have been sold only for research, not practical use. That makes recurring revenue questionable and the market dynamics look inflated with risk of a pullback.
688836.SS AVOID
NVIDIA locks in massive OpenAI infrastructure revenue.
NVIDIA is investing $100 billion in OpenAI's Ohio data center, which could scale to 8 gigawatts and generate an estimated $200-300 billion of revenue for NVIDIA from one customer alone. The contract also includes renewals for newer-generation GPUs, making NVIDIA a backstop for the AI buildout.
NVDA LONG
HIGH
14:02
Aug 20
Aug 20
Bankless
2d
GOOGL
MSFT
AAPL
▾
HIGH
Google/Microsoft legacy distribution moat persists.
Google and Microsoft have a legacy distribution moat even though they are not frontier AI model leaders: Microsoft is embedded across Fortune 500 companies through Teams and the Microsoft Suite, while Google is embedded through Google Docs and Google Suite, so enterprises and startups will keep clicking the convenient AI button. Josh added that Google seems slightly ahead of Microsoft but both benefit from this older customer base.
GOOGL WATCH
MSFT WATCH
Apple local AI could own consumer AI.
Apple is the AI wildcard for local, consumer AI: Apple's upcoming on-device model and event can make Siri pre-downloaded and competent across its massive installed base, eliminating the friction of downloading open-source weights; Josh argues Apple will own that entire local AI world and calls it the biggest event ever.
AAPL WATCH
HIGH
14:20
Aug 19
Aug 19
Bankless
3d
AI middleware/infrastructure layer
STRIPE 1ST
SPCX
MSFT FLIP
▾
HIGH
Value shifting to AI middleware layer.
AI value is shifting away from the model layer toward the middlemen: routing, inference, code hosting, and agent infrastructure. Stripe bought OpenRouter to own the meter where intelligence is consumed and priced, and Cursor/Origin is similarly positioning as a high-throughput tollbooth; the middle layer is where value should accrue next.
AI middleware/infrastructure layer LONG
Stripe becoming AI economic infrastructure.
Stripe is building the economic infrastructure for AI. Ejaaz argues inference has become a more valuable moat than training, and OpenRouter gives Stripe the routing/metering layer where intelligence is consumed and priced at a 5.5% take rate. Combined with Stripe's stablecoin, blockchain, and wallet efforts, it positions Stripe for agent-dominated microtransactions; he would love to own part of Stripe as the number one AI finance lab.
STRIPE LONG
SpaceX integrated AI coding stack leads.
SpaceX/Cursor is a frontier AI lab with a fully integrated software stack: models, Cursor coding/routing, and new Origin code hosting. Origin is engineered specifically for high-throughput AI agents rather than human commit speeds, giving it an advantage over GitHub and a tight feedback loop for future coding models. Investors can get public exposure through SpaceX stock, and Ejaaz is incredibly bullish on the whole setup.
SPCX LONG
Microsoft fumbling AI despite enterprise moat.
Microsoft has fumbled its AI opportunity: it owns GitHub and OpenAI IP but hasn't built a competitive AI product, Copilot is weak, and GitHub is now stressed by AI-driven commit volume and a serious Cursor/Origin competitor. Microsoft may be reduced to a cloud provider; its enterprise software/data moat and cash balance buy time, but the AI setup is unattractive.
MSFT AVOID
HIGH
13:52
Aug 18
Aug 18
Bankless
4d
GEV 1ST
INTC
MU
TSM
SPCX
▾
HIGH
Power and optics interconnect trade.
Gavin Baker's biggest new focus is power and optics: as GPU counts explode, moving data between GPUs over copper becomes inefficient, and photonics/light-based interconnects become cheaper and faster. Coherent is a major photonics manufacturer up 85% year to date, GE Vernova provides related power/material exposure, and Astera Labs supplies interconnect plumbing between GPUs, making this a consensus infrastructure trade.
GEV LONG
ALAB LONG
COHR LONG
Intel CPUs orchestrate AI GPUs.
Nvidia's $30 billion Intel stake makes sense because Intel builds bleeding-edge CPUs needed to orchestrate hundreds of thousands of GPUs and ensure AI agents make efficient, fast tool calls. Nvidia invested alongside the US government, and the CPU role in AI infrastructure is becoming more important.
INTC LONG
Memory supply sold out into 2028.
Memory remains a strong AI-infrastructure trade despite the Leopold blowup: SK hynix just reported record quarterly earnings but fell about 20% on fears that GPU/memory spending had peaked, while major investors doubled down at the lows. Ejaaz argues memory is physically supply-constrained, next year's memory supply is completely sold out, and demand extends into 2028, supporting Micron and SK hynix.
MU LONG
KS LONG
TSMC is AI picks and shovels.
TSMC is the picks-and-shovels AI-chip manufacturer: someone has to fabricate the chips, and TSMC is dominant at it. It is one of the most popular consensus positions among major funds, alongside Amazon and SpaceX.
TSM LONG
SpaceX is AI data-center winner.
SpaceX is a consensus AI-infrastructure winner because it is expected to be one of the most valuable companies in the world, will buy Nvidia GPUs to power data centers, has Grok reaching the frontier, and is building out data center capacity. Gavin Baker's $4.7 billion SpaceX position was cited as a big win.
SPCX LONG
Google owns the full AI stack.
Google is the only company that owns or dominates the entire AI stack, including models, custom TPUs, distribution platforms like Search, Gmail, G Suite and Android, plus infrastructure and power. Berkshire added roughly $17 billion to Alphabet, and Google's capex spending is on track for around $250 billion this year with strong cloud margins and strong capital-return history.
GOOGL LONG
SanDisk's HBF backlog extends to 2027.
SanDisk has developed and dominates a new memory category called high bandwidth flash, aimed at inference workloads that are consuming hyperscaler capex. SanDisk already has orders or backlog through the end of 2027, creating a new revenue opportunity with unprecedented excess demand.
SNDK LONG
QQQ puts hedge AI book.
Gavin Baker holds a $2.3 billion QQQ put position as explicit portfolio insurance. The hedge is designed to protect the fund if the broad market drops while it remains concentrated in AI hardware and infrastructure.
QQQ SHORT
Payment rails distribute AI inference.
Ejaaz argues AI plumbing will require payment rails to turn token spending into business intelligence. Visa and Mastercard are already integrated into companies' financial books and payment flows, potentially letting them distribute AI inference. He notes Stripe's acquisition of OpenRouter and frames this as an early, tinfoil-hat thesis but says the rails are positioned for it.
MA WATCH
V WATCH
AWS capex expands Amazon margins.
Amazon is a core AI winner because AWS under Andy Jassy is investing roughly $200 billion in AI capex while publicly reporting expanding AWS revenue and margins. Jassy ran AWS before becoming Amazon CEO and knows the infrastructure better than anyone, so Amazon is well positioned even as some funds trimmed and others bought.
AMZN LONG
HIGH
14:34
Aug 14
Aug 14
Bankless
8d
NEBIUS
SPCX
CoreWeave
INTC
AAPL FLIP
▾
HIGH
Nebius is high-growth, cash-rich neocloud
Nebius is a more mature neocloud with revenue up 454% year-over-year, compounding about 50% quarter-over-quarter, and about $8B in cash. It builds custom GPU clusters for specific workloads, which positions it to do well as AI demand grows.
NEBIUS LONG
SpaceX is back at AI frontier cheaper
Grok 4.6 puts SpaceX/Grok back at the AI frontier because it matches GPT-5.6 Sol and Fable 5 on intelligence benchmarks but is 3-5x cheaper to run. SpaceX has proprietary data, massive compute, and more powerful models coming soon. The stock was up 26% in five days because the market was not pricing this execution, and buying SpaceX stock is buying rocket, satellite connectivity, AI, and social media exposure.
SPCX LONG
CoreWeave backlog and GPU pricing are strong
CoreWeave had blowout results driven by GPU rental rates rising rather than depreciating. It has a $104B revenue backlog, $25B in Q3 customer commitments, NVIDIA backing and Anthropic contracts. The debt is high, but aggressive AI capex is the neocloud game to compete with hyperscalers.
CoreWeave LONG
Intel's raise signals foundry customer wins
Intel's upsized $20B equity raise after seeing $100B in demand is a clear signal that Intel Foundry likely has signed customers. Management had said they would raise external capital only if things were going well, so this points to a major custom-chip client such as Apple, Anthropic or OpenAI. Intel Foundry is positioned to compete with NVIDIA and AMD.
INTC LONG
GrokBot directly attacks Apple Siri AI
GrokBot is a direct competitive attack on Apple's upcoming Siri AI. GrokBot is an easy-to-use agentic mobile/desktop product aimed at knowledge workers, while Apple's Siri AI is not yet released and is constrained by Apple's closed ecosystem. This could pressure Apple's AI offering.
AAPL AVOID
HIGH
14:24
Aug 13
Aug 13
Bankless
9d
CoreWeave 1ST
NVDA
APO 1ST
SMH
GOOG
▾
HIGH
CoreWeave revenue surge shows GPU demand.
CoreWeave reported a 464% YoY revenue increase and signed an A100 contract extending to 2029, despite the 2020 GPU originally being expected to last 3.5 years. This shows versatile NVIDIA GPU demand and re-signing at higher prices; as a neocloud, CoreWeave is positioned to benefit from NVIDIA financing/revenue-sharing structures.
CoreWeave LONG
NVIDIA GPU supply tight and financing tailwind.
NVIDIA has a monopolistic, versatile GPU platform that even custom chips from Google, Meta, OpenAI, or Anthropic cannot match. The $500B GPU-financing deal turns GPUs into an asset class: NVIDIA provides depreciation protection/backstops up to 25% of opportunities, can take revenue splits from AI labs, and sees 2020-era A100s re-signed into 2029 at higher prices. Supply is sold out into 2027/2028 while AI demand compounds faster, so Ejaaz says he struggles to see a world where AI does not need NVIDIA GPUs and calls NVIDIA very bullish.
NVDA LONG
Financing partners win from GPU-backed lending.
Josh argues the assembled Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR can lend against transferable NVIDIA GPUs, collect high rates, and rely on NVIDIA's depreciation insurance/backstop; if borrowers fail, GPUs can be clawed back and redeployed. He calls the deal a huge win for the banks/asset managers.
APO LONG
BLK LONG
GS LONG
BX LONG
KKR LONG
BEP LONG
AI semiconductor supply remains constrained years.
Physical semiconductor supply cannot match AI GPU demand. Top memory manufacturers can expand capacity roughly 20% per year, while demand is compounding 45-60% per year, implying a constrained semiconductor supply chain until at least 2028/2029. That tight supply underpins pricing and supports the AI semiconductor complex.
SMH LONG
Watch Google capex return on AI.
Josh says the most important risk to monitor is hyperscaler return on invested capital. Using Google's earnings as the visible gauge, if AI spend stops generating revenue on schedule and margins shrink, hyperscalers will cut capex; if returns hold, the cycle remains fine. This makes Google a key signal for the AI infrastructure trade.
GOOG WATCH
TSMC wafer capacity constrains GPU supply.
Ejaaz names TSMC and wafer capacity as one of the hard limits on how quickly NVIDIA GPUs can be produced, which counters fears that new NVIDIA releases will flood the market and supports foundry pricing/volume tailwinds.
TSM LONG
HIGH
14:20
Aug 12
Aug 12
Bankless
10d
META
▾
MED
Meta's open-source AI pivot is bullish.
Meta's pivot back to open-source AI with models like Muse Glimmer, its aggressive compute capex, and its dominant AI glasses position leverage its 4 billion users to commoditize frontier intelligence, improve ad personalization, and unlock long-term value, making the stock attractive.
META LONG
MED
15:39
Aug 11
Aug 11
Bankless
11d
CRWD 1ST
PANW 1ST
▾
MED
Cybersecurity giants benefit from AI attacks.
The growing frequency and sophistication of AI-powered cyberattacks will drive massive demand for cybersecurity defense, directly benefiting companies like CrowdStrike and Palo Alto Networks, which have already posted record quarters as defense becomes a critical problem.
CRWD LONG
PANW LONG
MED
11:00
Aug 07
Aug 07
Bankless
15d
META
GOOG
▾
MED
Meta cheap AI, compute advantage bullish.
Meta's AI models are cheap and competitive on cost-per-token, and with its vast compute resources and aggressive expansion, it will capture market share from more expensive models; compute scaling laws favor Meta eventually becoming a top AI competitor.
META LONG
Google talent exodus, bearish short-term.
Google has lost its top AI talent, including Demis Hassabis stepping down as CEO of DeepMind and Jeff Dean leaving, casting doubt on its AI competitiveness; short-term bearish on Google stock.
GOOG SHORT
MED
09:07
Aug 06
Aug 06
Bankless
16d
SPCX
NVDA
▾
HIGH
SpaceX long-term bull case dominates
SpaceX reported 92% YoY revenue growth, Starlink doubled to 12M subscribers, and the company guided to $100B ARR by 2026. Its data centers pay back invested capital in 12 months, it has a $50B revenue backlog, and it is building a cloud compute business that sells AI capacity to hyperscalers. Despite short-term $100B share unlock overhang, the company has multiple shots on goal (connectivity, AI compute, space) with a path to $1T revenue by 2030, making the long-term risk/reward asymmetrically bullish.
SPCX LONG
Nvidia benefits enormously from SpaceX exclusivity
Elon Musk announced SpaceX will build its AI compute exclusively on Nvidia and the Vera Rubin architecture, aiming to double its GPU fleet by end of next year. This commitment could put $100-200B directly into Nvidia's pocket from SpaceX alone, making SpaceX potentially Nvidia's largest customer and a massive revenue driver.
NVDA LONG
HIGH
11:00
Aug 04
Aug 04
Bankless
18d
000660.KS 1ST
BE
IRON 1ST
NEBIUS 1ST
▾
HIGH
AI infrastructure demand remains strong.
Leopold Aschenbrenner's AI physical infrastructure thesis remains intact; hyperscaler capex and cloud revenue growth continue to surge, driving demand for memory, GPUs, and power infrastructure. The recent forced liquidation was due to excessive leverage, not a deterioration of fundamentals, and the sharp rebound in these stocks post-liquidation supports the view that the underlying demand story is strong, making current levels a buying opportunity.
000660.KS LONG
BE LONG
IRON LONG
NEBIUS LONG
HIGH
13:26
Jul 31
Jul 31
Bankless
22d
NVDA
MSFT FLIP
AMD
KS 1ST
INTC
▾
HIGH
GPUs will only go up in value.
The value of high-quality AI intelligence per token is going to keep going up, and because GPUs are the scarce resource that runs that intelligence, the value of GPUs will also only go up. Locking in rental compute now is a good idea because supply is limited while demand for intelligence is potentially infinite, driving GPU prices higher.
NVDA LONG
Microsoft cloud and AI business booming.
Microsoft reported $90 billion in revenue, beat earnings estimates handily, and its Azure cloud revenue grew 43% year-over-year, driving a 15% stock jump. AI demand is boosting cloud businesses and profit margins look impressive.
MSFT LONG
CPU demand from AI agents benefits Intel/AMD.
AI agents will require a lot of CPU infrastructure, where Intel and AMD are the number one providers. Both companies reported a strong quarter, profit margins are rising because of AI, and they are well positioned even though the stocks have been volatile.
AMD LONG
INTC LONG
Samsung memory demand locked in long-term.
Samsung's memory business is doing extremely well despite the stock dumping. They are signing long-term agreements with customers, effectively locking in memory supply out to 2027–2028, which signals very strong future demand.
KS LONG
Apple dominates on-device AI inference future.
Apple became the world's most valuable company, Siri is set to become incredible, upcoming devices like the iPhone Fold will be amazing, and Apple is in a uniquely strong position for running local AI inference on-device, setting it up for continued growth.
AAPL LONG
HIGH
14:52
Jul 30
Jul 30
Bankless
23d
000660.KS
ASML 1ST
SMH 1ST
▾
HIGH
SK Hynix selloff is a massive overreaction.
SK Hynix is the dominant HBM memory provider and just reported record earnings with 250% revenue growth and 600% operating profit growth, yet the stock sold off because it missed an arbitrary analyst revenue target by $1.7 billion. The miss was due to being completely sold out of HBM supply for the quarter, not weak demand. The market’s reaction is an overreaction; the underlying AI memory demand is exponential and insatiable, making the current dip a buying opportunity.
000660.KS LONG
ASML selloff on China news is overreaction.
ASML holds a unique monopoly on EUV lithography machines critical for advanced chipmaking. Recent news of a Chinese company replicating a DUV machine sparked an AI stock selloff, but the Chinese prototype is far from scalable production (only 5–10 machines planned) and poses no near-term competitive threat. The market panic is an overreaction, and ASML’s long-term demand remains strong.
ASML LONG
Semiconductors remain the core AI play.
The massive CapEx spending by hyperscalers (Google, Meta, Amazon, etc.) is flowing directly into semiconductor companies, creating an ‘unbundling’ of the AI trade. Semiconductor earnings are surging, and the demand curve for chips and memory shows no sign of slowing. The sector remains the best way to play AI infrastructure growth, and the recent selloff is unjustified by fundamentals.
SMH LONG
HIGH
13:57
Jul 29
Jul 29
Bankless
24d
SPCX
▾
HIGH
SpaceX mispriced, AI and compute drive value.
SpaceX’s rapid iterative manufacturing and compounding hardware efficiency (Starship reuse, Starlink V3, Colossus data center builds) give it an insurmountable competitive edge in both aerospace and AI infrastructure. The exponential improvement curve, analogous to Tesla’s 2018 breakout, makes long-term holding compelling despite near-term valuation concerns and supply unlocks.
SPCX LONG
HIGH
14:38
Jul 24
Jul 24
Bankless
29d
NVDA FLIP
GOOGL
TSLA
▾
HIGH
Vera Rubin chips lock NVIDIA’s dominance.
NVIDIA's new Vera Rubin chips deliver 10x efficiency over prior generation, drastically cutting AI model inference and training costs. This cements NVIDIA’s monopoly-like position; competitors like Google’s TPUs cannot catch up, and insatiable demand for GPUs will drive massive revenue growth and further stock appreciation. The chip's high-volume production ramp (1,000 racks per day) underscores the scale of the opportunity.
NVDA LONG
Google’s AI products are failing to deliver.
Google’s new flagship model Gemini 3.6 Flash is inferior to far cheaper Chinese open-source models, and the company hasn’t delivered a frontier model in months. Internal compute allocation is disorganized, and its next-gen chip Frozen V2 won’t arrive until 2028, putting it years behind NVIDIA. Despite record current earnings, investors are penalizing the stock (down 5% on the day of recording) because Google’s future AI competitiveness is in serious doubt.
GOOGL AVOID
Tesla’s decentralized AI network is the real bull case.
Tesla’s biggest bull case is its emerging decentralized AI network, built on cybercabs with integrated Starlink V5. These vehicles form a globally connected compute cluster for model training and inference, powering a new AI economy. Elon Musk’s track record in hyperscaling infrastructure and the near‑global internet coverage give Tesla a unique advantage beyond its car and robot businesses.
TSLA LONG
HIGH
14:46
Jul 22
Jul 22
Bankless
1mo
GEV 1ST
MU
000660.KS
▾
HIGH
GE Vernova leads AI power infrastructure.
GE Vernova is positioned at the center of the AI electricity trade, producing turbines and grid equipment. The stock has risen 300% in three years with orders doubling year-over-year to $7.1 billion in 2025 and backlog extending to 2031, offering highly predictable revenue. It has avoided the permitting issues that hit Bloom Energy, and has major deals with Microsoft and OpenAI. As data center power demand doubles in 24 months, GE Vernova is the old-staple, proven supplier poised for potentially exponential growth.
GEV LONG
Memory sell-off temporary, fundamentals still strong.
Memory stocks have sold off sharply (Micron down ~24% in a month) despite overwhelming demand fundamentals. DRAM prices are still rising (up 20% in July), and SK Hynix has already locked in 40% of its 2027 profits via long-term agreements with 13-15 customers, indicating sold-out supply. Micron trades at just 7x forward earnings with 350% revenue growth and 85% gross margins. The sell-off is driven by profit-taking and rotation into energy, not weakening demand. Oversupply is not expected until ~2030, making this pullback temporary and memory an attractive trade.
MU LONG
000660.KS LONG
HIGH
14:13
Jul 21
Jul 21
Bankless
1mo
Memory semiconductor stocks
NVDA
QQQ 1ST
▾
HIGH
Memory stocks benefit from growing AI demand
Memory stocks will be long-term winners because AI models need massive memory to store context about users, and infrastructure is the real moat. Cheaper models drive more usage, increasing memory demand. Recent price falls in memory stocks are temporary; the long-term trend is bullish.
Memory semiconductor stocks LONG
Open source AI spurs GPU demand, NVIDIA wins
Open source AI models like Kimi K3 increase demand for GPU compute, with NVIDIA being the main hardware beneficiary. Growing AI adoption and competition require more GPUs, independent of which labs lead, making NVIDIA a long-term winner. Current market dip is seen as a buying opportunity for GPU stocks.
NVDA LONG
AI sell-off overreaction; buy the dip
The market sell-off triggered by Kimi K3 is an overreaction because fundamental forces driving AI demand have not changed. US AI labs and infrastructure remain strong, making the current dip a decent buying opportunity for the broader tech market. Market resilience supports a long-term upward trend.
QQQ LONG
HIGH
15:34
Jul 16
Jul 16
Bankless
1mo
SpaceX AI
HBM 1ST
META
▾
HIGH
SpaceX AI's cheap models capture enterprise.
SpaceX AI's Grok 4.5 is a frontier-class model that uses 4x fewer tokens than Claude Opus 4.8, resulting in 17x lower cost for the same task. The model employs parallel agents, leveraging the Cursor acquisition for intelligent request routing. SpaceX AI has a fully integrated stack (data centers, custom chips) and an aggressive pipeline of larger models (5–10 trillion parameters). This enables them to deliver near-frontier intelligence at a fraction of the price, positioning them to dominate the price-sensitive enterprise market.
SpaceX AI LONG
HBM demand surges from AI chips.
Meta's new MTIA 400 chips use 51% more HBM, and the broader shift to custom AI silicon by hyperscalers will substantially increase demand for high-bandwidth memory. HBM is a direct beneficiary of the trend toward more memory-intensive AI accelerator designs.
HBM LONG
Meta wins low-cost AI with custom silicon.
Meta's MuseSpark 1.1 is the cheapest model per unit of intelligence, costing 25% of frontier models and undercutting even the cheapest Chinese open-source model. It uses a master agent for orchestration, is omni-modal (video, images, text), and is fine-tuned for computer use. Meta is vertically integrated with its own MTIA 400 silicon, making massive data-center investments. This shift to closed-source and the company's compute advantage position Meta as a serious competitor in low-cost enterprise AI, potentially capturing the broad middle market.
META LONG
HIGH
15:26
Jul 10
Jul 10
Bankless
1mo
NVDA FLIP
▾
MED
China chip push hurts NVIDIA's growth.
US export controls on AI chips have backfired, forcing Chinese AI labs like DeepSeek and Jipu to design their own custom ASIC chips. This reduces China’s reliance on NVIDIA GPUs, accelerates a siloed domestic hardware ecosystem, and threatens NVIDIA's future AI chip revenue from China. The gap between US and China is shrinking, and self-sufficiency could erode NVIDIA’s dominance in AI training hardware.
NVDA AVOID
MED
12:54
Jul 09
Jul 09
Bankless
1mo
DRAM basket
SKHY 1ST
005930.KS
000660.KS
MU
▾
HIGH
AI memory demand is overwhelming scarce supply.
AI training and inference require exponentially more memory than previous tech cycles, creating a structural demand black hole. Only three companies (Samsung, SK Hynix, Micron) can produce the high-bandwidth memory (HBM) needed, giving them monopoly-like pricing power. Supply is severely constrained: new fabs won't come online until 2030, HBM consumes 4x the wafer capacity of regular DRAM, and hyperscalers are begging for allocation. Profit margins are extraordinary (Samsung 52%, SK Hynix 72%). Despite recent share price drops of 20%, the fundamentals remain strong, forward PE ratios are low, and the sell-off is seen as a temporary 'sell the news' event. As long as AI usage keeps growing, memory demand will outstrip supply for years, making memory stocks a compelling long trade.
DRAM basket LONG
005930.KS LONG
000660.KS LONG
MU LONG
Buy SK Hynix US IPO as catalyst.
SK Hynix is listing on NASDAQ (ticker SKNY) on July 10th, raising $30 billion. The IPO is reportedly 4x oversubscribed, indicating huge institutional and retail demand. Ejaaz explicitly states he will buy the US listing, viewing it as a major catalyst that finally opens the leading HBM player to Western investors. The strong memory fundamentals and the company's dominant 60% HBM market share underpin a bullish entry point despite the broader memory stock pullback.
SKHY LONG
HIGH
14:36
Jul 08
Jul 08
Bankless
1mo
SPCX
TSLA
▾
MED
SpaceX leads space-based AI training
SpaceX is going to lead the emerging trend of training AI models in outer space using satellites with GPUs, which will become an increasingly important theme over the next couple of years and give SpaceX a dominant position.
SPCX LONG
Tesla Optimus scales like Model 3/Y
Tesla's Optimus humanoid robot program will leverage the company's proven ability to manufacture at massive scale (as seen with Model 3 and Model Y), making it a dominant player in the coming robotics wave.
TSLA LONG
MED
14:35
Jul 07
Jul 07
Bankless
1mo
CBRS 1ST
AVGO
NVDA
2454.TW 1ST
▾
HIGH
Cerebras is undervalued inference play.
Cerebras is a publicly traded inference chip company that the market is currently mispricing. The stock is down 35.5% from its IPO, despite inference demand exploding as AI agents and autonomous workloads drive token consumption. Ejaaz believes this creates an asymmetric bet, because inference performance is the new moat and Cerebras's accelerator chips are designed specifically for high-throughput, low-latency inference.
CBRS LONG
Broadcom offers asymmetric inference exposure.
Broadcom is a major designer of custom AI chips, including OpenAI's Jalapeno inference chip. Despite this, Broadcom is up less than 10% year-to-date, making it an undervalued play on the inference chip trend. Ejaaz frames it as another asymmetric bet in the inference space.
AVGO LONG
NVIDIA will adapt inference chips.
NVIDIA is aware of the shift toward inference and has acquired Grok, a custom accelerator company. Josh believes NVIDIA will integrate inference-optimized capabilities into future GPUs, creating a new product line. He warns against counting NVIDIA out, implying it will remain a leader even as custom ASICs emerge.
NVDA LONG
MediaTek benefits from inference chip design.
MediaTek is helping design custom inference chips for AI workloads and has already returned 180% year-to-date. Ejaaz includes it as part of the asymmetric opportunity in inference, implying that MediaTek's role in chip design will benefit from the shift toward specialized inference accelerators.
2454.TW LONG
HIGH
13:42
Jul 03
Jul 03
Bankless
1mo
META FLIP
MU
▾
HIGH
Meta GPU monetization is a smart stock driver.
Meta's pivot to selling cloud compute to third-party AI companies is a strategic, stock-positive move. Meta has accumulated one of the world's largest GPU fleets and can generate significant high-margin revenue by offering inference capacity, following the same playbook that SpaceX successfully used with Anthropic and Cursor. This monetization justifies Meta's continued heavy AI capex and was rewarded with a 6-8% stock rally.
META LONG
Memory sell-off overblown; scarcity persists.
The recent sell-off in memory stocks is an overreaction to speculative news about a breakthrough in memory efficiency. Memory is an incredibly constrained commodity with no scalable supply solution in sight; supply scarcity of advanced memory is likely to persist until at least 2028. Margins at memory producers like Micron have reached extraordinary levels, and the structural shortage is not resolved. The pullback creates a buying opportunity.
MU LONG
HIGH
14:52
Jul 01
Jul 01
Bankless
1mo
LITE 1ST
NEBIUS 1ST
MRVL
BE
▾
HIGH
Lumentum moves data as light, beats copper.
Lumentum provides optical fiber transceivers that move data as light, solving copper's heat and distance limitations inside AI data centers. Unit sales have roughly tripled, and NVIDIA invested $2 billion, signaling strong demand for photonics as clusters scale.
LITE LONG
Nebius rents scarce compute, revenue up 700%.
Nebius is a neocloud that rents out scarce AI compute at massive scale, with Q1 revenue up nearly 700%, NVIDIA backing of $2 billion, and multi-billion-dollar contracts from Microsoft and Meta. Its advanced permitting and regulatory approvals give it a strong competitive moat.
NEBIUS LONG
Marvell designs chip power delivery, Jensen endorsed.
Marvell Technology is the expert in designing the power delivery architecture inside AI chips, a critical and hard-to-solve part of the infrastructure. Jensen Huang publicly called it the next trillion-dollar company, the stock jumped 76% and entered the S&P 500, and it serves NVIDIA, Broadcom, and other AI chip designers.
MRVL LONG
Bloom Energy deploys power in 90 days.
Bloom Energy solves the AI data center power bottleneck with modular fuel cells that can be deployed on-site in 90 days, bypassing the 5-year lead time for grid connections or gas turbines. Revenue is up 130% year-over-year, and major customers like Oracle are using it for large-scale projects.
BE LONG
HIGH
14:18
Jun 26
Jun 26
Bankless
1mo
005930.KS
AVGO FLIP
MU
000660.KS
▾
HIGH
AI memory demand not peaked, suppliers rally.
AI memory demand is surging and far from peaked; Micron's blowout earnings and new Anthropic deal disprove the bear case that the memory trade is overcrowded. Only three major global memory suppliers exist (Micron, SK Hynix, Samsung), and they will all benefit as AI, autonomous vehicles, and robotics create exponentially growing memory needs that outpace model optimization.
005930.KS LONG
MU LONG
000660.KS LONG
Broadcom gains from custom AI chip trend.
Broadcom is a key enabler for hyperscaler custom AI chips. They are partnering with OpenAI to build the Jalepeño chip and already manufacture Google's TPUs, which the CEO says are on par with NVIDIA Blackwell. This positions Broadcom to rival NVIDIA in the custom AI accelerator market as more labs vertically integrate their hardware.
AVGO LONG
HIGH
14:48
Jun 25
Jun 25
Bankless
1mo
IBM 1ST
GOOGL
MSFT
QBTS 1ST
RGTI 1ST
▾
HIGH
IBM is a funded quantum leader with commercial traction
IBM has a unique position in quantum computing. The US government just allocated $1 billion (50% of its $2 billion quantum investment) to IBM. IBM Quantum already serves over 210 organizations with real-world applications (e.g., Cleveland Clinic for protein simulation, HSBC for trading algorithms). They are scaling qubits from 120 to 360 rapidly, and this early commercial traction combined with massive government backing makes IBM a compelling quantum play.
IBM LONG
Google and Microsoft lead quantum computing race
Google and Microsoft are the two clear leaders in fault-tolerant quantum computing. They have been investing for over a decade, possess the most resources, have published roadmaps showing orders-of-magnitude improvements every couple of years, and have already achieved triple-digit fault-tolerant qubits (Google Willow 105). As the field moves from R&D to real-world impact, these tech giants are best positioned to capture the next exponential wave, similar to how Google led in AI.
GOOGL LONG
MSFT LONG
D-Wave is an enterprise quantum optimization bet
D-Wave focuses on bridging quantum hardware and enterprise applications, optimizing the entire stack for practical use. It received a $100 million government check, is a small market-cap pure-play, and is trading with renewed volume since the Google Willow announcement. It offers exposure to the quantum adoption curve, though the technology is still pre-revenue and unproven at scale.
QBTS LONG
Rigetti is a small-cap quantum neocloud play
Rigetti operates as a 'neocloud' for quantum computing, owning the entire vertical stack from chip design to fabrication. It is a small-cap pure-play (~$7.1B market cap) that could capture quantum infrastructure growth. It was part of the US government's quantum investment and represents a high-upside, early-stage bet similar to early AI infrastructure companies, though it has not yet proven scalability.
RGTI LONG
HIGH
14:00
Jun 19
Jun 19
Bankless
2mo
GOOGL FLIP
META FLIP
▾
HIGH
Google losing AI talent is bearish.
Avoid GOOGL on weakening AI competitiveness; Noam Shazeer leaving Google DeepMind for OpenAI is framed as a bearish signal for Gemini/frontier AI talent retention, but no explicit short position, puts, or actionable short call is stated.
GOOGL AVOID
Meta's AI lead is gone.
Avoid META on AI execution concerns; speaker argues Meta AI has produced little despite heavy spending, reorgs, layoffs, and data-labeling pivots, but no explicit short position, puts, or actionable short call is stated.
META AVOID
HIGH
15:14
Jun 18
Jun 18
Bankless
2mo
GOOG
NVDA
SPCX 1ST
▾
HIGH
Google is quietly winning AI race
Google is underappreciated in the AI race. It is the only vertically integrated company that has mastered distribution and infrastructure. While missing from the top of the market-cap chart, it is 'very slept on' and will be a major competitor as frontier AI labs IPO.
GOOG LONG
NVIDIA more than just a GPU company
NVIDIA is underestimated as just a GPU play. Jensen has entered the CPU market, already generating $20B in annual revenue, and is building space-proof GPUs for Elon's rockets. This diversification and continued expansion beyond its core make it a resilient competitor.
NVDA LONG
Long-term SpaceX investors win
Josh is hyper-bullish on SpaceX for the long term, projecting it will become the most valuable company in the world. He cites Thomas LaFont’s probability research: companies reaching $1 trillion market cap have over 60% odds of reaching $10 trillion. He sees SpaceX's engineering team, vision (Starship, Starlink, AI compute), and leadership as unmatched, and a $10 trillion valuation as conservative if they hit $1 trillion in revenue by 2030. He recommends building a long-term position.
SPCX LONG
HIGH
14:36
Jun 17
Jun 17
Bankless
2mo
MRVL
BE 1ST
▾
MED
Marvell: AI infrastructure rotation winner
Marvell Technology sits at the intersection of optical fiber and power infrastructure, directly exposed to the next AI data-center buildout wave. Jensen Huang called it the next trillion-dollar company and NVIDIA invested $1.5 billion; Jensen's private investments have a strong track record of driving stock gains. Ejaaz is copy-trading this intersection as a beneficiary of rotating AI capital.
MRVL LONG
Bloom Energy hedged power-demand winner
Bloom Energy provides electricity throughput and benefits from unstoppable long-term energy and electricity demand that exists independently of AI. Even if AI demand slows, the general need for more power continues, making Bloom a hedged way to own the energy buildout thesis.
BE LONG
MED
15:47
Jun 12
Jun 12
Bankless
2mo
SPCX 1ST
AAPL 1ST
▾
HIGH
SpaceX’s IPO at a ~$2.1 trillion valuation is undervalued because the company has three massive growth pillars: Starlink’s explosive subscriber growth (12M+) and next-gen V3 satellites delivering 20x bandwidth per launch, AI data centers in space (AI-1 satellites with hot‑swappable GPUs, 150kW compute, laser‑interconnected grid, signed $56B/year deals with Anthropic, Google, Cursor), and unmatched hardware/manufacturing execution with fully reusable Starship bringing launch costs below $100/kg. Elon Musk’s proven ability to scale hardware from zero to one, the deep competitive moat (decade lead over Blue Origin), and the future merger potential with Tesla/Optimus/XAI make this a long‑term buy; the AI and space infrastructure bottleneck shifts to atoms, where SpaceX is uniquely positioned.
SPCX LONG
Apple Siri AI is a game-changer
Apple’s new Siri AI is a spectacular comeback; early beta reviews show it works faster and better than demoed, with true on‑device AI that is contextually aware of all personal data. Apple owns 3.5 billion devices and the custom silicon to run frontier models locally, giving it a unique advantage that no other AI lab can match—privacy‑first, no backdoor, no cloud dependency. This could subsidize and replace $20/month LLM subscriptions for casual users, making Apple a grand slam and very bullish long term.
AAPL LONG
HIGH
15:34
Jun 09
Jun 09
Bankless
2mo
AAPL
▾
MED
Apple owns consumer AI distribution.
Apple is well-positioned to dominate consumer AI because it controls 3.5 billion devices, has a vast moat of personal user data, and Siri AI now acts as a distribution funnel. The company can orchestrate its own foundational models or third-party models while maintaining privacy. Despite a 6% stock drop on WWDC day, the long-term investment case remains bullish as Apple owns the entire consumer AI distribution channel.
AAPL LONG
MED
13:45
Jun 05
Jun 05
Bankless
2mo
MRVL 1ST
NVDA FLIP
GOOGL
▾
HIGH
Marvell benefits from AI networking bottleneck.
Jensen Huang declared Marvell the next trillion-dollar company as it specializes in networking architecture for AI data centers, addressing the next bottleneck after compute and memory, and NVIDIA has invested $2 billion in the company.
MRVL LONG
NVIDIA expanding into CPUs, strong revenue.
NVIDIA announced entry into the CPU market at Computex, with a $20 billion revenue run rate expected this year, demonstrating successful expansion beyond GPUs into adjacent compute markets driven by AI demand.
NVDA LONG
Google's capital raise is bullish.
Google raised $85 billion in a strategic move to strengthen its balance sheet for future AI CapEx, which is bullish because they have $126 billion already and are using creative financial bookkeeping to avoid future dilution, expecting the stock to rise.
GOOGL LONG
HIGH
13:30
Jun 02
Jun 02
Bankless
2mo
DELL 1ST
▾
HIGH
Dell is a prime AI infrastructure play.
Dell is executing a successful pivot from a legacy server/PC company to a leading AI infrastructure provider. It assembles NVIDIA GPU racks and cooling systems, turning raw chips into functional AI data centers. The company reported record revenue with AI server sales up 800% year over year, has over 5,000 major customers, and was the first to deploy NVIDIA's Vera Rubin rack. The partnership with NVIDIA and the US onshoring push by the Trump administration provide strong demand tailwinds. Despite a massive stock rally, the fundamentals back the growth.
DELL LONG
HIGH
14:30
May 28
May 28
Bankless
2mo
MU 1ST
U 1ST
NVDA
ALAB
CBRS 1ST
▾
HIGH
Micron benefits from AI memory bottleneck
Micron is a leading memory manufacturer essential for AI workloads. Given the massive demand for memory chips from AI training and inference, and the company's recent 10x market cap growth to over $1 trillion, the memory bottleneck creates a huge opportunity for Micron.
MU LONG
Unity powers AI world model training
Unity Software is a world model builder with deep physics and 3D rendering capabilities. As AI moves toward AGI and humanoid robots, simulated environments for training become critical, and Unity's engine is one of the best platforms for creating these virtual datasets, making it a unique AI play.
U LONG
NVIDIA long, path to $10T
NVIDIA will maintain its high profit margins and strong demand as the AI supercycle unfolds, with a clear path to a $10 trillion market cap given its dominant GPU position and the insatiable demand from hyperscalers for training and inference compute.
NVDA LONG
Astera Labs solves AI connectivity bottleneck
Astera Labs provides the critical connectivity plumbing between GPUs in large-scale AI clusters. As data centers scale to hundreds of thousands of chips, the bottleneck shifts from GPUs to data transfer, and Astera Labs' solutions solve that, making it a key infrastructure play.
ALAB LONG
Cerebras captures inference compute growth
Cerebras designs specialized inference chips that are crucial as the AI industry shifts from pre-training to post-training and reasoning. Inference compute demand is estimated to be 5–10x larger than pre-training, and Cerebras is positioned to capture that growth alongside other inference-focused companies.
CBRS LONG
Bearish on broad market via QQQ puts
Gavin holds a large put position on QQQ (the Nasdaq 100 ETF) as a hedge against broad market downside. While he is bullish on specific AI infrastructure companies, he is bearish on the general market, believing the QQQ index may decline, and uses puts to express that view and protect against systemic risk.
QQQ SHORT
HIGH
13:00
May 22
May 22
Bankless
3mo
NVDA
SPACEX 1ST
▾
HIGH
Nvidia losing China GPU market
Nvidia has conceded China's AI chip market to Huawei after the US ban forced China to develop domestic alternatives. China now mandates that all superintelligent AI labs use Chinese GPUs, eliminating Nvidia's 20% profit contribution from China. This is a significant negative for Nvidia's revenue and competitive position, making the stock unattractive.
NVDA AVOID
Retail can buy SpaceX IPO shares
SpaceX is offering 30% of its share float to retail investors in its upcoming IPO, which is unprecedented and gives retail a rare opportunity to participate in the largest IPO in history. The company has strong revenue visibility from the Anthropic compute deal, and the valuation of $1.75 trillion is supported by future growth prospects including AI data centers in space.
SPACEX LONG
HIGH
Choose Sources
Loading sources...