
$NBIS estimated revenue growth is insane:
• 2025: $0.5B
• 2030: $33B (Est.)
That's 66x in 5 years at 129% annual growth rate.

• 2025: $0.5B
• 2030: $33B (Est.)
That's 66x in 5 years at 129% annual growth rate.
Layer 1: Materials & wafers
• $GLW
• $AXTI
• $IQE
• $AIXA
• $AMS
Layer 2: Core photonic devices
• $IPGP
• $COHR
• $LITE
• $LASR
• $SIVE
• Layer 3: Components & modules
• $AAOI
• $MTSI
• $FN
• $VIAV
• $LPTH
Layer 4: Systems & equipment
• $ASML
• $BESI
• $ASM
• $LPKF
• $MKS
Layer 5: Test, metrology & yield
• $CAMT
• $FORM
• $AEHR
• $ONTO
• $VIAV
$MU
Surged 29%+ in a single week, with market cap breaking through $800 billion — but the rally is far from over. HBM (High Bandwidth Memory) supply remains tight, and AI server expansion demand is pushing DRAM/NAND prices into an upward cycle. Conflicts in the Middle East have disrupted competitors' production schedules, making Micron the biggest beneficiary. HBM3E has already been adopted by NVIDIA and AMD, with analysts projecting related revenue to 2.5x by 2026, reaching $5 billion. Ahead of NVIDIA's May 20 earnings, the memory chip sector is likely to keep riding the wave.
$INTC
Driven by a dual catalyst — and the logic is rock solid. First: a preliminary agreement to manufacture for Apple, signaling that Intel's foundry business is being repriced by the market, reversing two years of negative narrative. Second: Q1 earnings beat expectations, with AI-related business accounting for 60% of revenue, and multiple institutions raising price targets. Already up 200%+ this year, but the re-rating of the foundry business is still in its early stages — Musk's companies are also planning to use its 14A process node. From a swing-trading perspective, there's room for a catch-up move after breaking previous highs.
$NVDA
The May 20 earnings report is the biggest near-term catalyst, and the market has entered "pre-earnings positioning mode." Year-to-date, it has significantly lagged the SOX index (only +15% vs SOX +61%), leaving clear room for a catch-up rally. Goldman Sachs has identified 5 potential upside surprises: upward revenue guidance, the new Vera CPU, sustainability of hyperscaler capex, growth among non-hyperscale customers, and gross margin resilience. Aggressive play: build a position 2 weeks before earnings, take profit or cut loss on the day results are announced.
$BIDU
The competitive landscape for domestic AI large models is stabilizing, Ernie Bot's commercialization is accelerating, and Baidu Cloud's AI revenue continues to grow strongly. Compared to the valuation froth in U.S. AI stocks, Baidu's AI assets are severely undervalued — and with earnings season approaching, the expectation gap is significant. Against a backdrop of marginally improving U.S.-China relations, overall sentiment toward Chinese ADRs is likely to recover. It edged up slightly last week and remains in a low-level accumulation phase — a good spot for swing-trade positioning.
$LI
Up over 2% last week, one of the few strong performers among Chinese ADRs. Li Auto has a clear positioning in China's EV competition (extended-range + premium family market), monthly deliveries continue to climb, and Q1 sales data was impressive. Domestic consumption policies keep being reinforced, and NEV subsidies are likely to continue, supporting overall sales volume. When sentiment toward Chinese ADRs warms up, Li Auto tends to have the most upside elasticity. Aggressive play: watch for pulse moves around monthly delivery data releases at the start of each month.
Small-Cap Picks
$ALAB
Focused on AI data center interconnect chips (PCIe Retimers, CXL, Ethernet), ALAB is a direct beneficiary of NVIDIA GPU server expansion, with products widely adopted by hyperscale cloud providers. Market cap in the $6–8 billion range — a true small-cap, high-growth name. As AI compute infrastructure upgrades toward super-node (128-GPU) architecture, demand for Astera's interconnect chips is exploding. Ahead of NVIDIA's earnings, sentiment across the entire AI supply chain is heating up, and ALAB's upside elasticity far exceeds NVDA itself. If earnings beat expectations, swing trade upside could reach 20–35%.
$SOUN
A representative small-cap AI application play, focused on voice AI platforms serving customers in fast food chains, automakers, healthcare, and more. Market cap around $1.5–2 billion — a classic "AI sentiment amplifier" that tends to move 3–5x the magnitude of large-cap leaders when AI enthusiasm heats up. Recently secured real deployment contracts with automakers and restaurant chains, transitioning from pure concept to actual revenue. With AI sector sentiment running hot, small-cap names like SOUN carry enormous upside elasticity. Aggressive play: follow the momentum when broad AI sentiment spikes — get in fast, get out fast, don't hold overnight positions too long.
$RKLB
Strong earnings last week (news specifically highlighted "Rocket Lab delivered outstanding earnings with several notable commercial milestones"), making it one of the few companies in commercial aerospace with a stable, recurring launch revenue base. Market cap around $4–6 billion — the most operationally substantive small commercial space company outside of SpaceX. Satellite internet, defense procurement, and the LEO constellation boom continue to drive demand, and RKLB is a direct beneficiary. A short-term post-earnings pullback is an entry opportunity, with a swing target at breaking previous highs. Compared to pure AI concept stocks, this one has real cash flow — relatively lower risk.
$MAAS
Originally a Chinese fintech and insurance agency company, MAAS has been aggressively acquiring and pivoting to AI since late 2025: completed the acquisition of Huazhi Future, gaining high-performance compute resources and AI algorithm R&D capabilities; plans to build the "Stars Distributed Intelligent Computing Center" with a total planned investment of RMB 5 billion; also acquired assets in new energy tech, mobile charging robots, and drinking water pipelines — the expansion logic is scattered, but the story is rich with themes. Stock price up ~63% over the past month, with Stocktwits retail sentiment at "extremely bullish." The AI computing center narrative resonates strongly with current market sentiment — there's still room for short-term pulse moves.
→ Jan 2025: $1B ARR
→ Dec 2025: $9B ARR
→ Apr 2026: $30B ARR
That’s a 30x in 15 months.
One analyst is now projecting $100B by end of 2026, $340B in 2027, and $2T+ by 2030.
Compare that to Google’s current revenue run-rate. The forecast says Anthropic could surpass it by mid-2028.
Is it too aggressive? Probably. But the direction of travel is real.
The bigger signal here isn’t Anthropic specifically — it’s what this means for the compute stack.
If AI model companies are monetizing this fast, demand for chips, memory, networking, power, and cooling is going to be far larger than the market priced in.
The infrastructure thesis just got stronger.
Before a potential Anthropic IPO, here’s where you can get exposure today:
→ $AMZN — lead cloud partner + investor
→ $GOOG — major backer + TPU development partner
→ $NVDA / $AMD / $AVGO — AI chip layer
→ $TSM — foundry capacity
→ $MU — HBM + DRAM demand surge
→ $MRVL / $FN / $LITE / $COHR — optical networking
→ $VRT / $MPWR — power & cooling
Pre-IPO fund exposure:
→ $VCX — Anthropic ~20.7% of portfolio
→ $DXYZ — meaningful Anthropic position
→ $AGIX — one of the few ETFs with direct private AI exposure
→ $BSTZ — private market tech exposure including Anthropic
The AI model race winner is still unknown.
The infrastructure winners are less uncertain.
$RKLB
— Only scaled launch alternative. SpaceX IPO validates the market, RKLB gets the multiple expansion.
$ASTS
— Highest-beta space name. Sector ETF inflows hit this first.
$PL
— Space data pure-play. IPO headlines bring retail back to Earth imaging.
$FLY
— Speculative flows lift all space-adjacent mobility names in a sector re-rating.
$LUNR
— Only pure-play Moon stock. Artemis ecosystem gets a spotlight when SpaceX lists.
$AXTI +612%
$SNDK +558%
$AEHR +382%
$AAOI +327%
$DOCN +241%
$INTC +238%
$BE +200%
$VIAV +187%
$WDC +179%
$MU +162%
$NVTS +155%
$LITE +145%
$OSS +118%
$AMD +113%
$NBIS +111%
$DRAM — +80.3% YTD
→ The pure-play memory ETF
→ SK Hynix 26% | $MU 25% | Samsung 21% | Others 28%
→ Maximum memory concentration in one ticker
→ Best if you want direct exposure to the DRAM/HBM supercycle
$EWY — +87.1% YTD
→ Plays the Korea memory duopoly — SK Hynix + Samsung in one fund
→ SK Hynix: 19%–24% | Samsung: 22% | Others: 54%–59%
→ Added macro beta on Korean won + broader Korean economy
→ Solid middle ground between pure memory and diversification
$KORU — +362.1% YTD
→ 3x leveraged version of $EWY essentially
→ Not a hold — a trade
→ Massive upside in a supercycle, brutal drawdowns on reversals
→ Only for those who understand leveraged ETF decay
$SMH — +52.6% YTD
→ Broader semi exposure with memory embedded→ $NVDA 16% | $TSMC 10% | $AVGO 7%| $INTC 5% | $AMD 5% | $MU included
→ Lowest memory concentration but highest diversification
→ Best entry point for those new to the theme
Risk Ladder (lowest → highest):$SMH → $DRAM → $EWY → $KORU
SPACE
$RKLB
$ASTS
MEMORY
$DRAM
$MU
$SNDK
COMPUTE
$AMD
$IREN
$NBIS
$INTC
+179%
Designs and manufactures CPUs for PCs and servers while rebuilding its foundry business to compete with TSMC.
$MU
+103%
Manufactures DRAM and NAND flash memory, the core storage components inside every server, PC, and smartphone, with demand accelerating sharply from AI workloads.
$ARM
+85%
Licenses the processor architecture that powers virtually every smartphone on Earth and a fast-growing share of data center and edge computing chips.
$MRVL
+84%
Develops custom AI networking and storage chips for cloud hyperscalers, becoming one of the most important picks-and-shovels names in the AI data center.
$AMD
+82%
Designs high-performance CPUs and GPUs for data centers, gaming, and PCs, and has taken meaningful share from Intel in servers over the last five years.
$AMAT
+52%
Supplies the deposition, etching, and inspection equipment that chip factories depend on to build every layer of a modern semiconductor.
$ASML
+31%
Makes the extreme ultraviolet lithography machines that are the only tools capable of printing the world's most advanced chips, giving it an unmatched monopoly position.
$TSM
+29%
The world's largest contract chip manufacturer, fabricating advanced processors for virtually every major chip designer including Apple, Nvidia, and AMD.
$AVGO
+18%
Builds custom AI accelerators, networking chips, and broadband semiconductors, supplying hyperscalers like Google and Meta with critical data center silicon.
$NVDA
+12%
Designs the world's leading GPUs for AI training, data centers, and gaming, making it the backbone of the modern AI infrastructure build-out.
Nvidia: 4m 23s
Microsoft: 5m 10s
Apple: 5m 36s
Amazon: 6m 46s
Meta: 8m 42s
• $NVDA owns main AI compute layer
• $AVGO monetizes custom silicon, networking & connectivity
• $AMD remains CPU bottleneck play as inference workloads scale
• $MU benefits from memory bottleneck across HBM & data center storage
PEG < 1 usually means mispriced growth
PEG > 2 starts to push into the danger zone
• Q1 2025: -$1,923M
• Q2 2025: -$23M
• Q3 2025: +$122M
• Q4 2025: +$803M
• Q1 2026: +3,615M
From -$1.9B to +$3.6B in just 5 quarters.
The top 3 memory gaints are all in there:
• $SSNLF (Samsung) - 25%
• $HXSCL (SK Hynix) - 25%
• $MU (Micron) - 24%