Why now is a good time to buy Intel

In ancient China, the "mandate of heaven" was the idea that a ruler governs with legitimacy granted from above. Intel's turnaround now has something similar: a leader chosen for the job, and explicit backing from Washington.

Here is the sequence of events, in order. Tap any event for the details and why it matters.

1. Mar 2025Lip-Bu Tan becomes CEO+

   Tan is a chip-industry veteran with 25+ years in chip design — he ran Cadence, whose software is used to design most of the world's chips — and a long career as a venture capitalist backing semiconductor startups. In other words: an insider who understands both the technology and the money.

   Why it matters: turnarounds are about the person at the top. Intel got someone who knows the whole chip ecosystem, not a career manager.

2. 2025 → 2026Engineers back in charge; layers of middle management cut+

   Since taking over, Tan has flattened the organization, removed layers of middle management, and pushed decision-making back toward the engineers building the products.

   Why it matters: Intel's problems were execution problems. Putting engineers back in charge is exactly what a chip company should do — and it's a very welcome sign.

   Source: CNN, June 7, 2026 ↗

3. Aug 2025The US government takes a 9.9% stake+

   The government converted CHIPS Act grants into roughly 433 million shares at $20.47 each (about $8.9B), becoming Intel's largest shareholder. The message was unmistakable: Intel is national infrastructure for the strategic chip needs of the decades ahead.

   Why it matters: that stake is now worth roughly five times what Washington paid — and it means the US government has every reason to keep Intel healthy.

   Source: Intel 2026 proxy statement (SEC) ↗

4. Jun 2026Former SK Hynix CEO joins to run advanced packaging+

   Intel hired Seok-Hee Lee — former CEO of SK Hynix, Korea's second-largest memory maker — as executive vice president of Intel Foundry, reporting directly to Tan. He leads advanced packaging and back-end manufacturing.

   Why it matters: you don't recruit the former CEO of a rival to run a side project. And in an August 2026 podcast Tan hinted that Intel is exploring new memory architectures — a possible comeback in a business Intel once led and left.

   Sources: The Korea Herald ↗ · Podcast recap ↗ · DigiTimes ↗

5. Jul 23, 2026Best quarter since 2011+

   Q2 revenue came in at $16.1B, up 25% year over year and well above Wall Street's ~$14.4B expectation. Data-center & AI revenue rose 59% to $6.3B; adjusted earnings per share were $0.42 versus $0.21 expected. Intel said it cannot keep up with data-center orders and has signed ten long-term supply contracts with server-chip buyers.

   Why it matters: the internal engine is improving — fast — while most of the stock story is still about the future.

   Sources: Intel press release ↗ · CNBC ↗

6. Aug 10–12, 2026$20B raised at $95 — about 5× oversubscribed+

   Intel set out to raise $15B; demand was so strong it upsized the deal to $20B, priced at $95 a share, with roughly $100B of orders. Big long-only funds and sovereign wealth funds led the way, and about a third of bidders received no shares at all. The proceeds fund an aggressive foundry expansion — 2026 factory spending has been raised to $20B+.

   Why it matters: sophisticated investors with full research desks looked at Intel near its all-time high and said: we want more, at this price.

   Sources: Intel newsroom ↗ · TechTimes ↗

7. Aug 11–14, 2026The CEO buys $10M of stock with his own money+

   Tan bought 105,263 shares at $95 (about $9.99M) in the offering, lifting his personal stake to roughly 1.33 million shares — worth more than $130M. The purchase was disclosed in an SEC filing on August 14.

   Why it matters: insider buying at an all-time high is rare. Executives usually buy after crashes, not after a stock has already tripled — unless they believe the run is far from over.

   Sources: TipRanks ↗ · TheStreet ↗

Three kinds of buyers, one direction

Everyone with the most information — the government, the biggest funds, and the man running the company — bought at prices within about 5% of today's, or far below it.

There are worries: interest rates, tariffs, the fear that AI spending is a bubble. I don't dismiss them. But they are not enough to bet against the biggest infrastructure investment of the 21st century — and two facts deserve more weight: how Intel is priced against its closest peer, and how large the demand wave in front of it is.

The peer comparison

TSMC (Taiwan Semiconductor) is the world's biggest chip manufacturer — it makes chips for Apple, Nvidia and almost everyone else. It's the closest thing Intel has to a yardstick. Compare the two on revenue and on market cap:

The only American company that can do this

A foundry is a factory that manufactures chips. Only three companies on Earth can make leading-edge chips at scale: TSMC (Taiwan), Samsung (Korea) and Intel (USA). If the United States is serious about the AI race — and it very much is right now — Intel is a strategic asset the country cannot afford to lose. The government's 9.9% stake says exactly that, and it will keep saying it for decades.

The size of the wave

The five biggest data-center builders in the US — Amazon, Alphabet (Google), Microsoft, Meta and Oracle — are on track to spend roughly $800 billion on data centers in 2026. Numbers that large lose meaning, so pick a comparison:

Even the cautious money is moving

Berkshire Hathaway — Warren Buffett's famously tech-shy conglomerate — has deployed a large slice of its cash into Alphabet (Google), one of the biggest AI spenders on the planet. Berkshire may not own Intel yet, but when the most conservative investors in America start betting on the buildout, that tells you the buildout is real. Business Insider ↗

And almost everyone is underestimating the market

Global chip sales rose 25.6% in 2025 to $791.7B. McKinsey argues the true size of the semiconductor industry is "hiding in plain sight" — far larger than the headline number once you count the compute embedded across the whole economy — and puts the race to build data centers at $7 trillion. Put simply: the market Intel serves is bigger than most people think, and growing faster than most people expect. McKinsey ↗

Hint: it's very simple. I asked Claude to explain scaling laws so that someone with zero AI background could understand. Here is the answer, unedited:

For most of AI's history, nobody knew how to make it reliably better. Progress came from clever new ideas, in fits and starts. You couldn't plan on it, and you certainly couldn't budget for it.

Scaling laws changed that. Around 2020, researchers found that AI gets better in a predictable way when you feed it three things: a bigger model, more data, and more computing power. Not sometimes. Predictably, smoothly enough that you could draw the line and forecast the result before spending the money.

That sounds dull, but it flipped the whole game. AI stopped being a lottery of clever ideas and became something more like a dial: turn up the resources, get a better model, on schedule. And when a result is predictable, people will invest in it. That's why companies suddenly felt confident pouring billions into chips, data centers, and electricity. They weren't gambling on a breakthrough; they were buying a known return.

It also explains why AI seemed to arrive out of nowhere. Once labs scaled models up dramatically, they didn't just get slightly better at old tasks. They started writing, coding, and reasoning in ways nobody explicitly programmed in.

So the current AI revolution in one sentence: someone found a dial that keeps working, and the world is now racing to turn it as far as it will go.

— Claude, asked by the author to explain scaling laws to a beginner

Remember: compute = chips. More GPUs, more compute. In even simpler terms: imagine a graph where the x-axis measures the amount of compute and the y-axis measures a model's "ability." AI researchers found a line that just keeps going up and to the right — the more compute you throw into training a model, the more capability seems to come out.

So the scaling law, in five words: more compute → better AI. And because compute is made of chips, the chain that drives Intel's future looks like this:

Where Intel fits in the chain

Nvidia gets the headlines for GPUs. But every AI data center also needs CPUs, needs its chips manufactured and packaged somewhere, and — if it's in America — needs an American manufacturer. That is where Intel sits.

Why now

Great investments usually happen when several things line up at once — and they rarely stay lined up for long. Right now, four things are pointing the same way.