Solid print from $GOOG yesterday. Search turned out stronger than expected, Cloud’s growth also not disappointing. Very underloved asset that has a stronger moat than most think imo, continues to be a comfortable core long.

This portfolio is up 40% since 04/12/24, outperforming the SPY (+4.4%) and NDX (+7.8%). That said, I think the portfolio wasn’t very well-designed as it didn’t consider cross-correlation and factor. We also haven’t screened volatility-adj returns so Sharpe might be lower due to the tariff episode.

It’s time to rotate out of a few positions that have played out sooner than expected and are now screening as expensive—NTDOY, HOOD, and COIN. We also want to increase exposure to major themes: AI (both U.S. and China), value-added software in the AI stack, rate-sensitive beneficiaries, robotics/autonomy, drones, and crypto.

Revised Allocation
60% Long-term Compounders: CDNS, GOOG, TCOM, NET, BABA

30% <1 year Event Trades: GLXY, AVAV, 002837.SZ, TER, BLDR

10% Moonshots: U, OUST

It remains a concentrated book; we want positions to matter when they work. But cross-correlation is also a consideration as the pf is meant to be hands-free: names must have distinct drivers to ensure diversification at the factor level

Closed the $BTBT short hedge at $3.3 yesterday — looking to reshort higher

Also trimmed more $8473.JT long @ $6033 Yen today

Added $INTC @ $23.05 — taking the print risk on this. Seems like most people are derisking into print. I could very well be wrong here but I think R/R is good enough for me

Added $AEO @ $11.8. Small position. Will cut quickly if the momentum dies down as retailers are too stable and slow for me.

Stock is up because WSB caught wind of the Sydney Sweeney campaign. Of all of the retail speculative stuff recently, this one at least makes a little sense to me as 1) campaign might reflect mgmt’s shift away from woke advertising, 2) fashion retail trends can inflect quickly, and 3) NTM P/E is still relatively decent at 11x?

Wrong on this; this turnaround is difficult and will take time. Cutting for 6% loss and moving on — not worth the oppy cost to be stuck in this position for long.

On second thought, I’ll hold and wait for the open to decide. The position isn’t large, so oppy cost is manageable.

LBT is precisely the kind of disciplined operator Intel needs at this stage — a ruthless executor focused on capital discipline and operational reset. He’s taking decisive actions: (1) a 25% reduction in workforce, (2) no new tape-outs without direct approval, (3) scaling back facility investments, and (4) limiting advancement on new nodes unless there is clear customer volume commitments and economically viable returns.

On point (4), it also reads as a not-so-subtle signal to the administration and domestic customers: without additional support, Intel — arguably the U.S.’s only viable chance at leading-edge foundry — will not continue to invest aggressively at the frontier.

This quarter is undeniably weak. But candidly, it’s better to surface all the operational and financial skeletons now and reset investors’ expectations, rather than endure a slow drip of negative surprises each quarter. After years of chronic mismanagement, there’s no real path to sustainable recovery without a painful but necessary restructuring.

Expanding more on (4), INTC can actually be rather.. compelling — if you exclude the oppy cost on a stock that may take years to play out.

They’re telling you upfront that they’re either (1) fully exiting the leading edge foundry business if there’s no willing customers and it continues to make a loss, or (2) remain in foundry if customers come and they can generate a profit on that line.

Assuming (1) as the worst case and excluding the foundry side completely from valuation, INTC will then essentially become just a fabless company.

So if you value the company just based on its Products line, ‘24 OpInc ~$13b, EV ~$130b, EV/EBIT ~10x. Considering that Products is also losing share to AMD and trending down currently, so say NTM OpInc ~$10b, NTM EV/EBIT ~13x. Avg US fabless EV/EBIT is around 18-25x. Even trading at the low range is a ~40% gain (the argument would then be does Intel even deserve to trade at the low range 😂)

But you get the idea; conservatively, zero-ing foundry, and assuming they continue losing share in Products, and excluding all of the optionalities present within the stock (e.g. US govt support, or exiting the foundry and focusing on Products helps INTC to win back share, or winning a large enough customer that allows 14A to work), it still does seem… undervalued?

I think the problem with Intel is Pat was always too rosy-eyed and hence allowed the company to bloat up in cost, all while building empty facilities first and “hoping” for customers to come and fill them afterwards (attitude of “if you build it, demand will come eventually”). LBT is more decisive and realistic, and actually willing to cut what’s not working (I’m probably biased towards the guy if you can’t tell).

Ofc SOTP maths always sound glamorous on paper… how market perceives it is another thing altogether — as drawing the line on the foundry business is an incremental factor that the market has to consider now — but just saying that it isn’t entirely bad and in fact it can actually work well.

Added $INTC @ $20.94

$TER +15% to $104 after a beat-n-raise quarter. Added to stock @ $89.8 yesterday pre-print.

Stock is performing well post-call given mgmt's increased confidence for re-acceleration in 2H driven by AI-specific semi test — a stark contrast to Q1 where they withdrew FY guidance. For a cyclical name like this, everyone wants to front-run the cycle bottom; hence even a quarter that "isnt as bad" can lead to a strong 1D price reaction. Now what remains is to see if that is indeed the cycle bottom for semi test as their end-market exposure arent as attractive as say Advantest.

$CDNS also reported strong earnings, now at $368. TACO Trump's export controls was indeed a good place to buy. One of my highest conviction plays this cycle too. I think there is still growth ahead for this company given that they are actually a very strong AI beneficiary; AI is helping them drive up to 20% PPA gains for customers, which is actually crazy given how node transitions are only getting more expensive. Its not cheap but you're paying for quality here.

Also shifted all $SBET positions to $BMNR following the 30% dip after PIPE unlock. Right now, I think BMNR will likely emerge the winner in this ETH treasury race.

Last core earnings to watch for this week is $NET. I love the company; I think they're going to be very critical in the new AI-driven web as they are the gateway to 20% of ALL internet sites. The CEO's recent tweets also sound like it's going to be a blowout quarter — you probably wouldnt roast another company if you're not going to beat and raise.. right?).

But the stock is so expensive (NTM P/E is 236x) that I cant bring myself to make it an oversized position going into print as even a slight mistake can be fatal. I have a position, but will add post earnings if there's a sell-off. This company can really surprise in the long-term despite its sky-high valuations now.\

Some final touches; (1) Replacing GLXY with RDDT, (2) Switching TER into compounders, exchanging positions with TCOM, (3) Adding BMNR in moonshots

Revised Allocation
60% Long-term Compounders: CDNS, GOOG, TER, NET, BABA

30% <1 year Event Trades: RDDT, AVAV, 002837.SZ, TCOM, BLDR

10% Moonshots: U, OUST, BMNR

These mega-cap earnings are insane. $MSFT $META

We’re seeing the first innings of ROI on AI spend. It’s going to accelerate from here; everything is computer, everything is power law driven.

Random side note. Ever get that doom feeling when using AI, like all tasks can and will eventually be replaced by it..? 70% of my workflow already runs on some form of it, and everyday I’m finding new ways to increase that %. I wonder a lot about how society will look like in 5 to 10 years.

Opened a position in $RDDT at $147. Everyone’s chasing the next social media giant; since $META, it’s mostly been letdowns ($PINS and $SNAP). Reddit feels like the sleeper.

As AI floods the web with copy-of-a-copy sludge, real and original human conversations, content and niche communities will only get more valuable (same reason I like $NET). Nobody wants LLMs trained on content made by other LLMs in an endless loop.

The valuation is also not wild: ~42× NTM P/E for ~30% top-line growth, with a path to software-like margins if it matures.

Key risk going into print is DAU and engagement. Google’s AI Overviews could dent clickthroughs and a slowdown in these metrics might spark a pullback, but I’d be a buyer on dips post-print as it doesn’t change the long-term thesis.

$RDDT and $NET will report in the wee hours later. Guess I’m not getting any sleep tonight.

$RDDT: Monster quarter and firing on all cylinders
• DAU +21% y/y to 110.4M
• Revenue +78% y/y to $500M (ads +84%)
• Gross margin 90.8%
• Adj. EBITDA margin 33.4% (+>1900 bps y/y)
• 3Q’25 guide: beats across the board

Engagement & traffic: Both logged-in and logged-out DAUs are humming, undercutting worries that Google’s algo shifts would hurt inbound. Anecdotally, Google’s forum modules (below AI overviews) are now stacked with Reddit links.

Monetization: ARPU was the best in recent quarters across U.S. and Intl. It tops second-gen social ($PINS, $SNAP) and still has runway versus META (though not calling for parity).

Ads flywheel: Advertisers +50% y/y with gains in both pricing and impressions. I take this as evidence of rising share of ad wallets and solid ROI for advertisers from community-driven intent.

Strategy check: Mgmt is locked on (1) core product, (2) search, (3) international. I view it as the right levers to boost engagement, harden against external platform’s policy shifts, and push the next leg of ARPU/user growth.

Overall a really strong quarter with nothing much to dislike. If I were to nitpick though, “Other revenue”— data licensing revenue — decelerated and got little airtime during the call. But licensing deals are likely lumpy and fall under NDAs, so I’ll give them the benefit of the doubt here. Another concern is how much of this beat is already priced in given $META’s blowout quarter the day before, but qtr to qtr fluctuations don’t sway me that much given that this is a mid-term hold.

As long as the core business and platform’s engagement continues growing, the stronger and more valuable Reddit’s data corpus will become. And licensing should follow in time.

Added alot more $RDDT @ $183 - $188, and $NET @ $199.

I think as long as RDDT's user growth continues trending in the right direction, the business will be more than fine and I'll bid. The magical thing about how RDDT's platform is structured is that it can bring value for advertisers by making it super easy to target the exact niche communities that they want (which will likely already be more interested in your products) -> driving clickthrough rates and ROAS. This positive flywheel can be sustained as long as user growth and engagement is strong, and I think mgmt realizes this and is focused in the right areas.

On $NET, just acting according to pre-print plan (buy dips post-print). Expensive business but I think their moat is unparalleled and will only grow stronger in an AI-driven web.