I forgot a third factor for Intel, maybe it’s even the most relevant one.

I was always taught not to overweight management’s tone/sentiment/track record etc, but I’ve found the opposite to be true. Sometimes, mgmt might even make up half of the story.

From when it all started, to a liquid 4x that you could jam in SIZE. Intel was and still is my largest long position, going into the print. I’m not selling yet.

I view it as crazily undervalued a year ago, and more fairly priced now -- but with lots of upside remaining. We have a good entry so we have the luxury of holding and waiting for the rest to play out.

Tailwinds:
- CPU demand exploding from agentic AI and inference (doubled from 1:8 to 1:4 CPU/GPU ratio per LBT, but I think he is conservative)
- Intel, with their own foundry, is the only one with supply capacity to flex to meet CPU demand, unlike AMD/ARM downstream that depend on TSMC that has no capacity
- Oppy in custom silicon, again having its own foundry is a plus against Broadcom/Marvell
- Oppy in EMIB packaging trending well
- foundry yields progressing well and better than expectations, 18A will meet EOY targets by mid 2026
- 14A yields progressing even faster than 18A
- 14A, which sounds like it’s on, will accelerate investors’ excitement in 2027 and beyond
- Intel being able to hit 41% GM, WITHOUT its foundry breaking even, or any external foundry customers, gives much to dream about for a blue-sky scenario
- strategic & political advantage; leading edge fabs in secure locations are ever more impt in today’s climate
- investors’ sentiments changing & turnaround accelerating

It’s a perfect storm. I also changed my view of Pat Gelsinger. He walked so LBT could run. Without his extensive investments that nearly broke Intel (for which he was often ridiculed for), Intel would not have a chance to catch up now.

$INTC

INTEL: APPLE, INTEL HAVE REACHED AGREEMENT FOR INTEL TO MAKE CHIPS IN APPLE DEVICES

Lately, there’s been a lot of commentary arguing that semis are in a bubble, often drawing lazy comparisons to the 1999 dot-com era.

Take $SNDK as an example. Just six months ago, FY26 consensus EPS was only ~$11. Today, the consensus for the EPS of that same year is closer to $64, nearly 6x higher. FY27 has moved even more dramatically, from ~$15 to ~$169.

The stock’s move over the past year has been almost entirely earnings-driven. That earnings expansion, in turn, reflects the growing importance of memory in next-generation AI.

It was only in the past few weeks that we finally saw meaningful multiple expansion, as LTAs may have changed the market’s perception of memory’s cyclicality and its supposedly commoditized nature.

So to argue that the past year was simply a bubble because “the stock went up 10x” misses the point.

The simpler (and more boring) explanation is that consensus estimates were just far too low a year ago, and failed to account for an inflection in earnings power. The stock price a year ago, got the future wrong.

One caveat is that we are probably getting closer to peak earnings (finally). You have to decide what’s the through-cycle earnings for yourself, and what multiples you can ascribe to that normalised earnings power. That may help you to sell right

On $INTC, flows can be constructive especially from long-only funds.

A year ago, this was widely viewed as a value trap: a failed turnaround, structurally disadvantaged in a “GPU > CPU” world, capex-heavy to the point of bankruptcy, “IFS will never work”etc. In short, it was almost uninvestable for many LOs.

Today, the narrative has flipped. What was once seen as a failed turnaround is now being framed as a must-own strategic fab story, a credible recovery, CPU reasserting relevance versus GPU, and “what other customers will IFS win besides AAPL?”

I remember a LO PM at my previous firm explaining how these funds operate. At a basic level, they seek to outperform by overweighting or underweighting names within their benchmark index. When you are underweight a stock and it rallies, you underperform the index. That underperformance directly hurts your carry.

With INTC up 6x over the past year, the stock’s index weight has risen materially, meaning an underweight position now creates a larger and more visible drag. At a $633bn market cap, INTC has gone from a small source of discomfort to a massive thorn in their backs. The LOs capitulation can drive the stock higher

$VICR +50%, and is also a proxy play on $CRBS IPO this week; Cerebras is the anchor customer for $VICR’s 1st fab

Starting to see that there are really only two trades in the market: semis and ex-semis. Mostly everything ex-semi (consumers, software, whatever you have) just ends up as funding shorts for semis.

AI is a power law and amplifies winner-take-all dynamics wherever it’s applied. Ex-semis are locked in a tight arm race for innovation, compute and tokens, and the fastest 20% to see ROI on their AI investments will absorb the other 80%. By that logic, semis can go higher, as everything ex-semis pour in $$$ to be part of the top 20% of AI winners.

By that logic have started heavily shorting consumer discretionary two weeks ago

$CAVA
$SHAK
$CMG
$SBUX
$LULU
$NKE
$DASH
$BKNG
$FIVE (not bc of this thesis, but just because its a shitty biz)

Apparently, $VICR BOM in $CRBS is near $20k, close to $TSM. Almost 4x expectations of ~$5k

Also added $WOLF at $38 and $STM @ $63 to the 800V transition / high-voltage power long basket

https://x.com/deedydas/status/2055491938464489888?s=46

Society crossed the Rubicon with GPT. The concept of work, education, and even money, is inexorably changed.

Resized my $TSLA short @ $422 on this pop from $SPCX hype. Biggest short pos

Added long few weeks back; $NOK common @ $12.5 as third largest position, along with Jan’27 $25 calls

Valuation is reasonable, along with TAM expansion in scale-across from edge/inference demand explosion. Also has a fat right tail from physical AI providing growth optionality over the next 5-10 yrs

“UBS analyst Timothy Arcuri out on Micron $MU trying to take the place of Mary Meeker / Henry Blodget with his sky high price target of $1625 from $535

Valuation: PT to $1,625 (from $535) We raise our price target from $535 to $1,625 based on a ~15x NTM P/E multiple (noting we see no reason why MU should trade a whole lot differently than NVDA in terms of P/E), which is in line with the 3-yr average, and applied to our ~$117 C2029E EPS and discounted back to C2028E using a COE of ~12%. Importantly, we anchor on C2029E EPS as we believe it best reflects MU’s through-cycle earnings power under LTAs because, by that point, our model assumes a moderate memory downcycle, albeit with LTAs buoying the company’s earnings base.”

Here comes the multiple expansion… Been adding to my large mcap core positions ($INTC $MU $SNDK $LITE) over the past weeks, while the attention was all on small caps

A Trillion United States Dollars for Micron Technology $MU

Who’s next? Team blue?

Built up a mid-size $SMTC long position at $147 in commons and calls, which has exposure to both copper and optics. One of the stronger optics names past week. Earnings tonight

Power names have gone vertical as market starts taking the 800V transition more seriously

$VICR @ $340 from $200
$NVTS @ $32 from $14.5
$WOLF @ $75 from $38
$STM @ $71 from $63
$CORZ @ $26 from $19
Also added $ENPH @ $49, now $67.5

Very speculative basket as I'm optimising for torque here, so caution should be exercised. Despite this, the 800V transition and the obstacles that come with it, along with the companies positioned to solve it, are very much real. For blue chips, maybe $IFX and $ON.

Added a long on $PURR @ $9

- partially hedging out my crypto related shorts $COIN $CRCL
- Hyperliquid is probably the one of maybe 5 valuable crypto protocols to emerge in the past 5 years.
- 24/7 and pre-IPO perpetual markets is a major tailwind for HYPE given the numerous mega IPOs upcoming in 2026/27
- if DATs make a comeback, PURR has an add. benefit of mNAV expansion. Ex high finance mgmt seems to know how to optimise dilution too

Trimmed 10% of my memory exposure $SNDK $MU last Friday, and added more $LITE @ $830 during the dip. Also picked up $AXTI @ $115

Optics has chopped lately but nothing has changed about the thesis. After memory, interconnect is probably the next lowest hanging fruit

Google’s equity raise and Marvell’s Computex presentation this week should help refocus the market’s attention back on this theme

$SHAK guidance cut sending tremors across other fast casual / consumer discretionary

Just reported a month ago so it’s intg why they had to cut guidance in such a short timeframe. Consumer deterioration accelerating?