Richard Medina

Everyone is debating $NVDA vs $AVGO vs $AMD in the AI chip race. The real question is who wins regardless of the outcome. The answer is $TSM. CoWoS is the technology that bonds AI chips together with high bandwidth memory. Without it, the most powerful AI chips in the world cannot exist. Every single one runs through TSMC. Here is how 2026 capacity breaks down. Source: Morgan Stanley. $NVDA: 595,000 wafers. 60% of global demand. $AVGO: 150,000 wafers. 15%. Google TPUs. Meta. OpenAI. $AMD: 105,000 wafers. 11%. Three companies. Three different strategies. One factory. Nvidia is betting on general purpose GPUs. Broadcom is betting on custom silicon. AMD is fighting for enterprise data centers. They are all competing against each other. And they are all paying TSMC to do it. CoWoS is sold out through 2026 and into 2027. TSMC CEO C.C. Wei confirmed it. That is a bottleneck for $AVGO and $AMD delivering on their commitments. But that same constraint is exactly what makes $TSM the most defensible position in AI infrastructure. TSMC is targeting 130,000 CoWoS wafers per month by late 2026. Nearly 4x late 2024 levels. More capacity means more revenue. And the customers locking it in are the biggest companies on the planet. You can debate which AI chip strategy wins. You cannot debate who builds them all.​​​​​​​​​​​​​​​​ $TSM

@The_AI_Investor The Taiwan risk is real but that’s also our advantage since it trades at a lower valuation due to it. If the Taiwan risk comes true then all of Semiconductors are getting hit hard in the short term

$TSM wins regardless, but too bad they have the highest Taiwan risk, otherwise it's a no brainer. Most stocks are exposed as well at different levels of impact.

@danielisdizzy I think it’s that they will buy the compute from anywhere that can get it! There’s not enough compute capacity and their revenue is growing exponentially

There’s a growing rumor that Anthropic will strike a major deal with $AMD to use MI450 accelerators. Anthropic’s strategy is clear: They don’t want to depend on anyone.

@Gage769 @StockMKTNewz Seriously! It’s nice not having to think about the category

@RichardMInvest @StockMKTNewz 3% on all purchases kinda op

What credit cards do you have/use?

@SoCarTider @StockMKTNewz $50 charged annually

@RichardMInvest @StockMKTNewz What is annual fee?

@SRJonesLA @Leo_Traydes Came back strong this week! If the bull market continues we will be more than fine

@RichardMInvest @Leo_Traydes HOOD is killing me

If your portfolio isn’t at all time highs what are you holding? Genuinely curious

@JoshKale Anthropic is building the tools to be a full fledge operating system with their models as the engine. The pace of product launches continues to be outstanding 🚀

Three days ago, Anthropic's CPO left Figma's board. Today, Anthropic became Figma... Here's everything to know about Claude Design: → Describe what you need. Claude builds v1. You refine through chat, inline comments, direct edits, or custom sliders Claude generates on the fly. → Web capture tool: point Claude at your live site, pull real elements into prototypes so they look like the actual product. → Reads your codebase and design files during onboarding, then auto-applies your brand system to every project forever. → Turns static mockups into interactive prototypes. → Exports to Canva, PDF, PPTX, or standalone HTML. Ships designs as org-scoped URLs. → Handoff bundle → Claude Code → production. One instruction. → Pages that took 20+ prompts in other tools now take 2. Anthropic is taking no prisoners and Figma is down 83% on the year.

@oguzerkan Interesting how he mentions buying too much compute could bankrupt them and not enough will cause them to fall behind. The company that can forecast demand more efficiently will have the advantage.

Looks like Anthropic massively underestimated the compute they would need this year. Dario previously implied in a podcast that they were taking a more conservative approach as overestimation of demand could lead to bankruptcy. That turned out to be a strategic blunder. People are switching to ChatGPT because Claude’s limits are so low. Compute demand will reaccelerate from here. Anthropic will try to close the compute gap with OpenAI while OpenAI will try to widen it. Compute wars begin. Semis and neo-clouds will be the biggest beneficiaries, especially the neo-clouds that have already secured the power. $NBIS $IREN $ORCL $AMD

@StockSavvyShay Anthropic keeps launching new products at record speeds. They are becoming an operating system built on top of their models! Compute demand will continue exploding

Anthropic just introduced Claude Design which is a tool that lets users create prototypes, slides & one-pagers by talking to Claude. This starts pushing into territory usually associated with $FIG & $ADBE but through natural language & without needing much design experience.

@StockSavvyShay @FuturumEquities @SchwabNetwork $NVDA, $AVGO, $AMD all run through $TSM . The AI buildout is accelerating and TSMC just confirmed it. The next 5 years will blow out our expectations

$TSM just reminded everyone that the AI infrastructure cycle is still accelerating and that it remains the core factory behind the new AI economy. The first phase was about GPU but the next phase is about the full stack around them as the bottlenecks shift into foundry, packaging, memory, CPUs, networking & power.

@StockSavvyShay @brewmarkets Let’s all have Diamond hands and look back at our portfolios in 10 years 👀 This is a good reminder that patience pays off in high quality companies

A $10K investment in $MU at the IPO would be worth ~$3.2M today. Honestly a brutal reminder of what patience looks like when it actually works.

@Leo_Traydes I think profit taking! $TSM is increasing operating margins while increasing CAPEX. They will be supply constraint for years. The story is only getting better

Two powerhouses at the forefront of AI are struggling today post earnings… is this just profit taking? $ASML $TSM

Everyone is focused on the revenue beat. I am focused on this👇 $TSM operating margin trajectory : Q4 2024: 49.0% Q4 2025: 54.0% Q1 2026: 58% Straight up. Every quarter. And they did this while consistently growing capex. $40.9B in 2025. $52B to $56B in 2026. More capex means more capacity. More capacity means more AI chips. The concern was always that heavy spending would crush margins. Q1 2026 just proved the opposite. Spending more. Earning more efficiently. That is exactly the trend you want to see.

@StockSavvyShay Anthropic isn’t slowing down the product launches! Compute demand will keep going up. So bullish for $NVDA $TSM $AVGO

Anthropic launched Claude Opus 4.7 across all Claude products today. The new model is designed to improve software engineering, higher-resolution vision and instruction following.

Jensen Huang just admitted one of the biggest missed opportunities in tech history. When OpenAI and Anthropic needed billions to scale, $NVDA was not in a position to invest. So the labs went to Microsoft, Google, and Amazon instead. And in return committed to using their compute. Jensen’s words: “I always thought they could just go raise from VCs. But what they were trying to do couldn’t have been done through VCs. I recognize that now. I didn’t know it then.” That decision is why Anthropic runs on Google TPUs. Why OpenAI is deeply tied to Azure. Why hyperscalers now sit at the center of the AI supply chain. Not because they built the best models. Because they wrote the check when nobody else would. Jensen’s takeaway from missing it? “I’m not going to make that same mistake again.” Capital is strategy. Whoever funds the lab controls the compute relationship. The hyperscalers figured that out first. Nvidia is making sure it does not forget it. x.com/dwarkesh_sp/st…

$TSM beat every single line and guided higher. Down pre market anyway. Revenue: $35.9B. Beat ✅ EPS: $3.49. Beat ✅ Gross margin: 66.2%. Beat ✅ Q2 guidance: $39B to $40.2B. Above estimates ✅ The fundamentals just got stronger while the price got cheaper. Are you buying shares at these prices?

@wallstengine We need more capacity! Huge win for the AI buildout. More capacity = more revenue for all 🚀

TSMC SAYS CAPEX OVER THE NEXT 3 YEARS WILL BE “SIGNIFICANTLY HIGHER” THAN THE LAST 3 YEARS, WHEN IT SPENT $101B.

@wallstengine These #’s prove the AI cycle is far from over! Arizona Fab 2 confirmed on 3nm with CoWoS packaging coming. It will be the first complete AI chip supply chain on American soil. That is the real story today.​​​​​​​​​​​​​​​​ $TSM is reducing their geopolitical risk even further 🔥

$TSM | TSMC Q1’26 Earnings Highlights 🔹 Revenue: $35.90B (Est. $35.5B) 🟢; +40.6% YoY, +6.4% QoQ 🔹 Net Profit: $18.1B (Est. $17.0B) 🟢; +58.3% YoY 🔹 Gross Margin: 66.2% (Est. 64.5%) 🟢 🔹 Operating Margin: 58.1%; +9.6 pts YoY 🔹 Net Profit Margin: 50.5% 🔹 CapEx: $11.10B; +10% YoY 🔹 Wafer ASP: $7396; +10% YoY 🔹 EPS: $3.49 per ADR Segment Performance: 🔹 HPC: 61% of net revenue; +20% QoQ 🔹 Smartphone: 26% of net revenue; -11% QoQ 🔹 IoT: 6% of net revenue; +12% QoQ 🔹 Automotive: 4% of net revenue; -7% QoQ Other Metrics: 🔹 3nm: 25% of wafer revenue 🔹 5nm: 36% of wafer revenue 🔹 7nm: 13% of wafer revenue 🔹 Advanced technologies (7nm and below): 74% of wafer revenue 🔹 Wafer Shipments: 4,174K 12”-eq.; +28.1% YoY; +5.4% QoQ 🔹 North America: 76% of total net revenue Capital Return: 🔹 Dividend: NT$6.00 cash dividend for Q4’25