For most of this year, the semiconductor trade had one simple logic to it.

AI needs chips. Chips need fabs. TSMC builds the fabs. Buy TSMC.

Then this morning, TSMC reported Q2: $40.2 billion in revenue, up 36%. Net profit up 77.4% to a record. AI chips now 61% of revenue. Q3 guided to $44.6 to $45.8 billion.

The best single quarter in the history of semiconductor manufacturing.

And the obvious trade got louder than ever.

But the loudest trades on the loudest mornings are rarely the most disciplined ones.

This week's Pragmatic Playbook is on TSMC. The business is genuinely exceptional. And I have a specific problem with the question "should I jump on it?"

MarketBeat’s take is that Fastenal’s post-earnings drop looks more like disappointment over “only meeting expectations” than a real business problem. Revenue grew nearly 15%, daily sales rose 14.7%, net income grew 14.9%, and the company returned nearly 80% of net income through dividends and buybacks. The bigger message is that Fastenal’s digital inventory tools, like FastBin and FastVend, may be quietly strengthening its moat even if the stock pulled back.

The Fool’s argument is that Apple could overtake Nvidia again because two catalysts are lining up: stronger iPhone momentum in China and the launch of Apple Intelligence there after regulatory approval. Apple shares are already up 20% in 2026, its market cap is around $4.8 trillion, and the article says Chinese iPhone sales grew 24% year over year as higher memory costs pushed Android vendors to raise prices. In plain English, Apple may be benefiting from AI indirectly, not by selling chips, but by using its supply-chain strength and ecosystem to win share.

TSMC’s Q2 profit jumped 77% year over year to a record T$706.6 billion, comfortably beating expectations, as demand for advanced AI processors from customers like Nvidia and Apple stayed strong. The company also reported roughly $40 billion in revenue and record gross margin around 67.7%, but the stock still slipped as investors had already priced in a lot of good news. The takeaway is simple: the AI supply chain is still booming, but expectations for the biggest winners are now extremely high.

Scheduling meetings isn't why you get out of bed. Filing expenses isn't your calling. So hand it all to Catch — the AI admin that takes the tedious work off your plate and gets it done.

Just ask. Catch handles the back-and-forth on scheduling, flights, restaurants, follow-ups, vendors, clients and more — then comes back with it sorted, in moments.

No guesswork. No dropped balls. What's yours stays yours, private by default and reliable by design.

Nobody became a leader to book their own flights. Meet your admin savior and go from swamped to sorted.

Get started free at catchagent.ai — and start spending your time on the things only you can do.

Get Started Now

TSMC (TSM) delivered $40.2 billion in Q2 2026 revenue, up 33.7% year over year in US dollars. Net profit hit a record, up 77.4%. Gross margin of 67.7%. Q3 guided to $44.6 to $45.8 billion.

The easy trade: TSMC is the most irreplaceable company in AI infrastructure, and every advanced AI chip in production today runs through its fabs. Buy it.

But "should I jump on it?" is the wrong question.

The right question is what you are actually buying at $430, the morning after the best quarter in semiconductor history, when every investor is reading the same headline and arriving at the same conclusion.

Nobody makes these margins building physical objects.

A gross margin of 67.7% on $40.2 billion in quarterly revenue is, for a company operating around-the-clock semiconductor fabs, nearly not supposed to exist. Software businesses struggle to reach those numbers. TSMC delivers them while building the most complex physical objects humans have ever manufactured.

Advanced nodes, specifically 3nm and 5nm, account for 77% of wafer revenue. That is where the pricing power lives. As AI models grow larger and require denser chips, demand for TSMC's most advanced nodes increases, not decreases.

That is a margin expansion story dressed as an earnings report.

The Q3 guide of $44.6 to $45.8 billion tells you this was not a one-quarter anomaly. The full-year revenue growth outlook was raised to over 40%. Sequential acceleration from an already-record base.

Revenue up 36%. Profit up 77%. AI chips: 61%.

That last number is the one worth slowing down on.

AI and high-performance computing now represent 61% of TSMC's total revenue. Three years ago, TSMC's revenue was spread across smartphones, PCs, IoT, automotive, and servers. No single category dominated.

Today, TSMC is an AI infrastructure company that still makes some smartphone chips.

The businesses most exposed to an AI capex pause right now:

Not because TSMC is fragile. Because concentrated demand creates concentrated exposure. The 36% revenue growth works as long as Amazon, Microsoft, and Google keep spending at the rate they announced. If Amazon's Q2 print on July 30 introduces any capex hesitation, that conversation does not stay in Seattle. It lands directly on TSMC's Q3 orderbook.

That is the risk the headline does not show you.

TSMC committed $100 billion in additional US investment alongside these results.

And honestly? That is bullish information wrapped inside a risk acknowledgment.

Companies do not spend $100 billion to diversify production geography unless the risk of not doing so is real enough to justify the cost. The Taiwan geopolitical discount in TSM's stock has always existed because it should exist. TSMC spending $100 billion to partially reduce it is the clearest signal that its most important customers, Nvidia, Apple, AMD, and the hyperscalers, have committed enough future demand volume to justify Arizona-based advanced production.

That is a long-term vote of confidence from the customer list. It does not appear in the Q2 P&L yet. It will.

TSM hit its 52-week high of approximately $473.50 on June 22, then quietly gave back ground in the three weeks before this morning's print. The stock drifted from that peak into the $415 to $430 range not because anything changed in the business, but because institutional investors do not add size ahead of earnings at peak valuations when they can simply wait for the data to arrive.

Post-earnings today, the reaction has been measured rather than explosive. Record results, but the $60 to $64 billion capex commitment and H2 margin guidance are keeping buyers cautious. That is the market telling you the good news was already partially priced, and the elevated capex creates uncertainty about TSMC's 2026 free cash flow that the Q3 numbers will need to resolve.

The $410 level is the first meaningful support below current prices. A settlement below $420 in the coming sessions tells you to watch $410 carefully. A recovery and hold above $450 shifts the narrative and sets up a test of the June high at $473.50. Below $410, the next real floor is closer to $380.

TSMC's earnings trajectory is a derivative of hyperscaler AI spending. Watch the transcripts.

Because 61% AI revenue concentration means TSMC's Q3 and Q4 depend on Amazon, Microsoft, and Google maintaining their capex commitments. Amazon Q2 on July 30 is the first data point. If it confirms AWS revenue acceleration with no capex pullback, TSMC's Q3 guide is probably conservative. If any of the three signals capex hesitation, the concentration story looks different by October.

The Arizona fabs are not just a geopolitical hedge. They are a pricing catalyst.

Watch for any update on TSMC's N2 (2nm) Arizona production timeline. US-based advanced node production commands higher selling prices because elevated operating costs get passed through to customers as a margin premium. When Arizona N2 volume starts contributing meaningfully, TSMC's already-exceptional 67.7% gross margin gains another expansion lever that does not depend on Taiwan capacity or geopolitical risk pricing.

The $450 level is the technical tell before the next quarterly data arrives.

Watch whether TSM pushes above $450 and holds in the coming weeks. A stock that reclaims $450 after a record print tells you institutional buyers are adding conviction at the new earnings base. A stock that churns between $410 and $440 despite record results tells you the market is waiting for another quarter of data before committing size. Those are two setups, and only one of them is worth adding into before Q3 earnings.

The margin story depends entirely on TSMC maintaining its yield advantage at advanced nodes.

Watch for any credible update on Intel Foundry or Samsung's 2nm progress. The competitive picture has only concentrated further since then. Neither threatens TSMC in the near term. But any meaningful yield improvement at scale at a competing foundry changes the pricing power story underpinning the 67.7% gross margin.

TSMC reports monthly revenue, giving investors a real-time window into Q3 before the quarterly call.

Watch the July and August monthly prints. TSMC's June revenue was already up 68% year over year. Monthly data consistent with $14 to $15 billion per month confirms the Q3 guide is tracking. A miss against that pace gives the market something to reprice weeks before the Q3 call arrives.

The best quarters in company history are the worst times to make entry decisions. Not because the business changes. Because the crowd does.

TSMC is genuinely exceptional. A 77% profit jump, 67.7% gross margins, and a Q3 guide implying the record quarter was just the warm-up. The thesis is right: if AI wins, TSMC wins. It is the only company in the world that can manufacture the most advanced AI chips at production scale. Samsung has tried. Intel has tried. Neither can match TSMC's yield rates at 3nm and below.

The problem is not the business. The problem is the verb.

"Jumping on" means reacting to a headline. It means buying at $430 the morning the best quarterly print in semiconductor history lands in every investor's inbox simultaneously, when the stock is already up 90% from where it traded twelve months ago.

That is not an entry. That is a reaction dressed as a decision.

The investors who will do best with TSMC from here are not the ones who jumped on it this morning. They are the ones who already owned it at $250 and are adding on the first real pullback below $410. Or the ones who did not own it, can be patient, and wait to buy with conviction rather than urgency.

TSMC is not going anywhere. Neither is the AI capex cycle it sits at the center of. The entry discipline is what compounds over five years. Not the same-morning reaction to the best quarter ever.

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide.

The best quarters in history are the hardest times to make disciplined decisions.

Every data point confirmed the thesis. Every headline pointed the same direction.

That is the moment to slow down, not speed up.

TSMC's business did not become irreplaceable this morning. It has been irreplaceable for years. What changed today is that everyone else found out at the same time.

Conviction is built in the research, not in the headline.

Know the difference going into the weekend.

If this week's Playbook kept you from a reactive entry, forward it to one friend who needed the same lens.