💻 McKinsey Sees a $2.3 Trillion Semiconductor Market by 2030: Where Will AI Create the Most Value?

Hey Tigers 🐯!

The AI semiconductor story may be much bigger than previously expected.

In its latest outlook, McKinsey raised its 2030 base-case forecast for the global semiconductor market to $2.3 trillion.

Its previous estimate, published in late 2025, was:

$1.6T → $2.3T by 2030 📈

That is a 44% upward revision in less than a year.

But for investors, the bigger question is not simply:

“How large can the semiconductor market become?”

The more important question is:

Where is that additional value being created — and which parts of the semiconductor supply chain are benefiting most from the AI infrastructure boom?

🚀 AI Is Changing the Semiconductor Growth Curve

The semiconductor industry has always been cyclical.

But the current AI cycle looks different.

According to McKinsey, semiconductor sales growth in the first half of 2026 reached its highest level since the mid-1980s, while industry revenue surpassed:

$1 trillion for the first time

McKinsey now expects the semiconductor market to grow at roughly:

19% CAGR through 2030

For comparison:

Previous forecast → 13% CAGR

2014–2024 historical growth → ~9% CAGR

That is a significant acceleration.

And one major force is driving it:

AI infrastructure spending 🤖

Five major hyperscalers have announced approximately:

$800B CapEx in 2026

$1T CapEx in 2027

A large share of that spending is expected to support AI infrastructure.

More AI data centers mean more demand across the entire semiconductor ecosystem:

• AI accelerators

• Advanced logic chips

• HBM and memory

• Networking

• Advanced packaging

• Power infrastructure

But there is another important difference from previous semiconductor cycles.

AI is not only driving higher chip volumes.

For some of the industry's most valuable chips, it is also supporting:

Higher demand + Higher pricing 📈

That combination is one reason the current semiconductor cycle looks so different from a traditional PC or smartphone-led recovery.

The key takeaway:

AI is not simply creating another semiconductor demand cycle — it may be changing the industry's growth curve itself.

🏗️ Data Centers Are Becoming the Growth Engine

🏗️ Data Centers Are Becoming the Growth Engine

This is where the numbers get especially interesting.

McKinsey expects semiconductor revenue tied to servers and data centers to grow from:

$330B in 2025 → $1.2T by 2030

That implies roughly:

29% CAGR 📈

The scale of that growth is significant.

By 2030, the data-center semiconductor segment alone could become larger than what many investors once expected the entire semiconductor market to be worth.

Meanwhile, other end markets are expected to grow much more slowly.

For example:

Wireless semiconductors → ~10% annual growth

The contrast is clear.

AI-related infrastructure is becoming a much larger part of the semiconductor growth story, while traditional consumer electronics are playing a smaller relative role.

The key takeaway:

The semiconductor market is increasingly being driven by AI infrastructure rather than traditional consumer-device cycles.

That means the next phase of industry growth may depend less on smartphone or PC replacement cycles — and more on how quickly hyperscalers continue building AI data centers.

🧠 Which Segments Could Capture the Most Value?

Not every semiconductor segment benefits equally from this AI cycle.

McKinsey identifies two major areas as the key engines of value creation:

⚙️ Leading-edge logic chips

Advanced logic used in AI accelerators and high-performance computing could capture a major share of industry growth.

💾 Memory — especially HBM

As AI models become larger and more data-intensive, high-bandwidth memory is becoming increasingly critical to overall system performance.

Together, these two categories could account for a large share of the semiconductor industry's incremental value creation through 2030.

🧠 Where Is the Value Being Created?

Between 2025 and 2030, McKinsey estimates that two areas could drive a large share of semiconductor value creation:

⚙️ Leading-edge logic: +$710B

Advanced chips used in AI and high-performance computing are expected to be one of the biggest growth engines.

💾 Memory: +$560B

Memory — especially HBM — could add another major layer of value as AI workloads become more memory-intensive.

What makes this cycle unusual is not just the growth in demand.

It is the pricing.

Normally, as a manufacturing node matures, semiconductor prices gradually decline.

But today, McKinsey notes that prices for 3nm and 5nm wafers are rising by 2%+ per year.

Why?

Because demand for AI and high-performance computing remains strong, while advanced manufacturing capacity is still limited.

That creates a rare semiconductor setup:

Higher volume + Higher average selling prices 📈

For investors, this helps explain why the AI semiconductor opportunity has moved far beyond GPUs alone.

The broader AI infrastructure cycle includes:

• Advanced foundries

• HBM and memory

• Advanced packaging

• AI accelerators

• Networking infrastructure

Different companies sit at different points in this value chain.

For example:

$NVIDIA(NVDA)$→ AI accelerators
$$Taiwan Semiconductor Manufacturin $Taiwan Semiconductor Manufacturing(TSM)$ → advanced foundry capacity
$$Micron Technology $Micron Technology(MU)$→ memory and HBM
$$Broadcom $Broadcom(AVGO)$→ AI networking and custom silicon

The key takeaway:

AI semiconductor growth is increasingly becoming an ecosystem story, not just a GPU story.

💾 HBM Is Becoming One of AI’s Biggest Bottlenecks

GPUs usually get most of the attention in AI investing.

But increasingly, memory bandwidth is becoming just as important as compute power.

Large AI models need enormous amounts of memory to:

• Store model parameters

• Move data between processors

• Support larger context windows

• Handle more tokens

• Serve more users at the same time

That makes HBM — High Bandwidth Memory — a critical part of modern AI accelerators.

As AI models become larger and more demanding, the pressure on memory systems keeps increasing.

McKinsey expects memory supply to remain relatively tight in the near term.

Additional capacity and improvements in advanced packaging could allow supply to begin catching up with demand around 2027.

But even if pricing eventually normalizes, McKinsey still expects memory to remain one of the semiconductor industry's largest sources of value creation through 2030.

For investors, this changes the way the AI chip opportunity should be viewed.

It is no longer simply:

“Which GPU company wins?”

Instead, the opportunity is increasingly spread across the entire AI infrastructure stack.

GPU → HBM → Foundry → Packaging → Networking → Data Center

That is why the AI semiconductor theme is becoming more of a:

System-level infrastructure trade 🚀

rather than a single-stock GPU trade.

⚠️ But the $2.3T Forecast Comes With Conditions

⚠️ A $2.3T Market Is Not Guaranteed

McKinsey’s $2.3 trillion semiconductor market forecast is a base case — not a certainty.

For that scenario to play out, several things still need to go right.

1️⃣ Data-center CapEx must stay strong

The world may need roughly:

$2.7 trillion in cumulative data-center investment through 2030

That means hyperscalers cannot materially slow their AI infrastructure spending.

If CapEx from companies like Microsoft, Alphabet, Amazon and Meta starts to weaken, semiconductor demand could cool much faster than expected.

2️⃣ AI needs to generate real returns

Infrastructure spending can only continue if AI eventually creates measurable economic value.

Investors should watch whether enterprises are actually seeing:

• Higher productivity

• Lower operating costs

• New revenue opportunities

• Stronger margins

If AI ROI disappoints, companies may become more cautious about infrastructure spending.

3️⃣ Power could become a major bottleneck ⚡

AI data centers consume enormous amounts of electricity.

That means future AI expansion increasingly depends on:

Power generation → Grid capacity → Data-center construction

Even if semiconductor demand remains strong, insufficient electricity infrastructure could slow deployment.

4️⃣ Supply needs to stay relatively tight

McKinsey expects shortages in leading-edge chips and HBM to potentially persist for another:

3–5 years

That tight supply environment helps support strong pricing and margins.

But there is also a risk on the other side.

Today’s shortage could become tomorrow’s oversupply.

If manufacturers add too much capacity too quickly, the industry could eventually return to a more traditional semiconductor downcycle.

👀 What Should Investors Watch Next?

Instead of focusing only on semiconductor revenue, these indicators may tell us more about where the cycle is heading.

💰 Hyperscaler CapEx

Are $Medifirst Solutions Inc.(MFST)$ , $Alphabet(GOOGL)$, $Amazon.com(AMZN)$ and Meta still increasing AI infrastructure budgets?

Continued spending would support demand across GPUs, ASICs, HBM, foundries and networking.

📈 AI ROI

Are companies actually generating productivity improvements or new revenue from AI?

This could become one of the most important indicators for whether AI infrastructure spending remains sustainable.

🧠 HBM & Advanced-node Capacity

How quickly can the industry expand supply?

If capacity remains tight, pricing power could stay strong.

If supply catches up too quickly, margins could come under pressure.

💵 Chip Pricing

Can advanced chips and memory maintain elevated average selling prices?

One unusual feature of the current AI cycle is that both:

Volume ↑ + Pricing ↑

are happening at the same time.

Whether that continues matters enormously for semiconductor earnings.

⚡ Power & Data-center Infrastructure

Can electricity generation, grid capacity and data-center construction keep up with AI demand?

Increasingly, the bottleneck may not be computing demand itself.

It may be the infrastructure required to support it.

🌎 The Bigger Picture

McKinsey’s forecast is about more than a larger semiconductor market.

The bigger question is whether AI is changing the economics of the semiconductor industry itself.

Historically, semiconductor growth usually looked like this:

Higher chip volumes → Lower prices over time

AI is temporarily changing that relationship.

Today, demand for advanced computing and memory is rising so quickly that the industry can experience:

Higher volume + Higher pricing

at the same time.

If that dynamic continues, the semiconductor opportunity could be substantially larger than investors expected even a year ago.

And that changes the key investment question.

Instead of simply asking:

“Which company has the best AI chip?”

Investors may increasingly need to ask:

“Where is the next bottleneck in the AI infrastructure stack?”

🐯 What are you watching most closely?

GPUs, ASICs, HBM, foundries, advanced packaging — or data-center infrastructure?