To The Moon

Last Thursday looked ugly for semiconductors.

AMD fell 3.9% to $620.68. Nvidia lost 2.94%, Broadcom dropped 4.35%, while Intel fell 5.34%.

AMD then weakened again on Friday, closing around $608.

Normally, when stocks fall this hard, I ask whether the fundamental story changed.

This time, Lisa Su gave us an unusually useful answer.

AMD is planning a substantial increase in supply, yet management still sees demand exceeding what the industry can currently produce. Advanced wafers, HBM, packaging and other parts of the AI supply chain remain constrained.

That creates a fascinating divergence:

The stocks are pricing fear while the supply chain is still pricing scarcity.

For me, that is not automatically a sell signal.

It is a signal to become much more selective.

Why Did Chips Sell Off Then?

The market is no longer debating whether AI demand exists.

It is debating whether companies can earn enough from the enormous amount of money being spent to justify current valuations.

That is a much harder question.

We now have several forces hitting AI stocks simultaneously:

High Treasury yields

Expensive valuations

Enormous AI capital requirements

Questions about AI customer economics

Oil and inflation pressure

Profit-taking after extraordinary semiconductor gains

This is why good fundamental news can coexist with falling stock prices.

The market is moving from:

Is AI real?

to:

What return will all this AI spending generate?

That transition creates volatility.

But AMD Just Gave Us an Important Signal

Lisa Su is not describing a market struggling to find customers.

AMD is trying to secure more capacity.

That distinction matters enormously.

The company is working across TSMC, packaging partners and memory suppliers because the constraint is still how much hardware the ecosystem can produce.

AMD is also planning capacity several years ahead.

To me, that does not sound like management preparing for an imminent collapse in AI demand.

It sounds like a company trying to secure enough supply to participate in a market that remains capacity constrained.

The Bottleneck Is the Thesis

AI accelerators are not built from GPUs alone.

They require:

Advanced wafers.

HBM.

Advanced packaging.

Substrates.

Networking.

Power.

Cooling.

Data-centre capacity.

That means the AI boom increasingly resembles a chain of bottlenecks.

Fix one and another appears.

This is why I remain constructive on selected semiconductor companies even while the market questions AI spending.

If demand genuinely exceeds supply, the companies controlling scarce capacity still have pricing power and visibility.

The question is which companies capture the economics.

AMD Is Becoming More Interesting After the Selloff

AMD traded above $658 earlier this week before closing Friday near $608.

That is already an approximately 8% reset from the week's high.

I would not call $608 cheap.

AMD has had an enormous run and expectations remain demanding.

But I like it considerably more at $608 than at $650+.

The next question is whether buyers defend the psychological $600 area.

If they do, I think AMD can begin rebuilding.

If $600 fails decisively, the correction probably has another leg.

AMD Pick Levels

$600-$610: First Pick

We are essentially here.

I would use this for a starter position, not full size.

$570-$590: Preferred Pick

This is where the risk/reward becomes much more attractive to me.

If AMD reaches this zone because Nasdaq and semiconductors continue correcting while AI demand remains intact, I would add.

$530-$560: Strong Pick

This would represent a meaningful reset.

I would become substantially more aggressive here if the fundamental AI demand story has not changed.

$630-$650: First Recovery Confirmation

AMD needs to reclaim this region before I trust another momentum leg.

Above $660: Breakout

A decisive move above the recent high would tell me buyers have regained control.

$700-$725: First Bull Target

If AMD breaks $660 while supply-demand commentary remains strong, this becomes my next target area.

$750-$800: Bull Extension

Possible, but I would need continued earnings revisions and stronger evidence that AMD is converting AI demand into earnings at scale.

Below $530: Thesis Check

At that point I would investigate whether the market is pricing something more fundamental than valuation compression.

Nvidia Is Still the Benchmark

I would not interpret AMD strength as Nvidia weakness.

AI compute demand can be large enough for both companies to win.

Nvidia remains the dominant AI accelerator ecosystem.

AMD represents the challenger with potentially greater share-gain asymmetry.

That makes their risk/reward different.

If AI infrastructure spending remains enormous, Nvidia does not need AMD to fail.

And AMD does not need Nvidia to collapse.

AMD simply needs to capture a larger percentage of a rapidly expanding market.

That is why I find AMD particularly interesting on corrections.

Broadcom Is the Other Name I Would Watch

AVGO falling to around $360 also gets my attention.

Broadcom gives investors another route into AI infrastructure through custom accelerators and networking.

Its question is different from AMD.

For AMD, I ask:

How much AI accelerator share can it capture?

For Broadcom:

Can custom AI silicon scale fast enough that profit growth outruns margin pressure?

I like AVGO around $345-$355 for a first pick and $325-$340 as my preferred accumulation area.

I would not chase it back toward $380 without confirmation.

Intel Is the Most Complicated

Intel falling harder than the others makes sense to me.

AMD and Nvidia are fundamentally AI demand stories.

Broadcom is an AI custom-silicon and networking story.

Intel is still partly a turnaround story.

Terafab creates enormous optionality, but optionality is not the same as earnings.

I therefore demand a much larger margin of safety.

Around $107, Intel is interesting.

Around $100-$105, I become more interested.

Around $90-$98, I would view the risk/reward much more seriously if the foundry and Terafab thesis remains intact.

But Intel remains the highest execution-risk trade of these four.

What I Think the Market Is Missing

There is a subtle difference between:

AI demand slowing

and

AI stocks becoming too expensive.

Right now I think the market is primarily testing the second.

That is healthy.

If AMD management were saying customers were cancelling orders, HBM availability was loosening rapidly and wafer capacity was suddenly abundant, I would react very differently.

Instead, management is trying to secure more supply.

That tells me the physical economy of AI still looks considerably stronger than Thursday's stock charts suggest.

My 2-4 Week AMD Scenarios

50% Base Case: $580-$650

AMD digests its enormous run, tests $600 and trades sideways while investors debate AI valuations.

This is my highest-probability scenario.

I accumulate selectively rather than chase.

30% Bull Case: $650-$700

The semiconductor selloff stabilises, Treasury yields ease and AMD reclaims $650.

A break above roughly $660 could bring momentum buyers back quickly.

20% Bear Case: $530-$580

High yields and AI financing concerns continue pressuring technology.

AMD breaks $600 and undergoes a deeper valuation reset.

Ironically, if Lisa Su's demand thesis remains intact, this could create the best buying opportunity of the three scenarios.

What Would Make Me More Bullish?

I want several pieces to line up.

AMD holds $600.

Treasury yields stop climbing.

HBM remains constrained.

Advanced packaging remains tight.

Hyperscaler AI capex stays strong.

AMD continues securing capacity.

And most importantly, AI revenue converts into stronger earnings and free cash flow.

If AMD then reclaims $650-$660, I become much more constructive.

What Would Make Me Bearish?

Not another 5% red day.

I would become concerned if the physical supply-demand evidence changes.

Customer orders weaken.

Capacity suddenly becomes easy to obtain.

HBM shortages disappear much earlier than expected.

Hyperscalers cut AI capex.

AI customers struggle to finance infrastructure.

Or AMD expands capacity aggressively just as demand rolls over.

That last scenario is the real danger.

The worst semiconductor cycles usually begin when enormous supply arrives just as demand slows.

We are not there yet.

My Pick

Of the four names in this selloff, AMD is my preferred aggressive pick.

My map:

$600-$610: First Pick

$570-$590: Preferred Pick

$530-$560: Strong Pick

$630-$650: Recovery

Above $660: Breakout

$700-$725: First Target

$750-$800: Bull Extension

Below $530: Thesis Check

I would also watch AVGO on further weakness, while NVDA remains the quality benchmark and INTC remains the higher-risk turnaround.

The market is currently asking whether AI companies are spending too much.

Lisa Su is telling us something different:

AMD still cannot secure enough supply to satisfy the demand it sees.

Both statements can be true.

That is precisely why I do not want to chase semiconductor rallies, but I also do not want to panic when they correct.

The best opportunities may emerge when stock prices are falling faster than the underlying demand is.

Right now, AMD is getting close to that territory.

I am not a financial advisor. Trade wisely, Comrades!

Which would you buy on this semiconductor reset: AMD, NVDA, AVGO or INTC?

If you find this post interesting, give it wings! 🪽 Repost and share the insights.

Do consider ‘Follow me’ and get firsthand read of my daily new post. Thank you. 

@Tiger_SG  @TigerEvents  @TigerStars  @TigerPM  @Daily_Discussion