Preface: I publish my portfolio update every ~3 months to my family. I commend the monthly portfolio posters to Saul’s board. It really is a lot of work and I read each one in their entirety. My portfolio update is not exactly in the Saul-like format. I include some market opinion assessments as education to the family who do not follow market news much. I encourage them to buy non-Saul ‘foundation’ safer stocks. But I expose my aggressive portfolio activities to try and teach. So here it is:
Zane’s Q3 2026 Portfolio Update for Family
The S&P500 is already up 14% in only 6 months. AI is driving much of this. Examples are chip companies, memory companies, Neoclouds (Nebius) and the Data Center connectivity/AI infra players. Hyperscalars like Microsoft, Amazon, Google, Meta are spending like drunk sailors on AI data centers. Historically, they have been cash cows building equity every year with high profit margins. Now they are spending their huge cash reserves. More cash is going out than coming in. (too much capex?). The future return on investment is fuzzy but clearer for some companies. So, the challenge is to pick the winners this year. Market skepticism and political headwinds create short-term risk, but constructive price action and software sector resilience support a bullish outlook
IMO the AI boom is not an immediate bubble. Rather it is a structural move to a massive productivity economic boom. All the major hyperscaler companies know this as well as the major bankers on Wall Street. We may very well get a bubble a couple years when the demand weakens. The time is not now. Right now, we will see data centers pop up everywhere. A lot of jobs are being created. There is a political movement against these data centers partly based on unfounded fear by doomers and partly because communities Do need to control these changes they will live with. Regulation at the local level is necessary. Regulation at the federal level is very murky and doubtful, though you will hear a lot of political rhetoric. The fear of AI may create some ambiguity in the market and create some buying opportunities. But the AI revolution is foundational to the US economy and the multi-layered network of businesses. IMO there is no stopping the AI momentum and rational regulation is ambiguous. It reminds me of when McDonalds, WalMart, Costco, and Home Depot moved into communities and destroyed the local businesses with lower prices. But new jobs were created and better goods and services emerged. This is capitalism. You’re seeing incredible compute demand because of it. In this new world of AI, compute is revenue.
In 2026 my Fintech/Payments/Crypto Financial stocks have pulled hard back because of fear and uncertainty due to the war, tariffs, and the macro economy. Two Fed rate cuts were expected in 2026. Now none are expected with rising inflation and perhaps an increase. This means there is not much stimulation to drive more borrowing which drives the banks and fintech earnings. I am holding onto partial investments patiently even with substantial losses in 2026. The pressure is on SOFI, NU, HOOD, and FIGR is macro. These companies continue to perform on all fundamental profit metrics. I have long term profits on the first three, so it is an easier boring hold. But when I see a good opportunity, I find myself each time selling a little bit of these stocks.
Foundation stocks. Notice that I do not personally hold many foundation stocks as I live on a very volatile investment train. This risk is not what I recommend to family members that are concerned with capital retention.
Strong Buy stocks now: High Quality Foundation stocks. My preference Buys are NVDA, AMZN, META, MSFT, and NU. BTW I purchased META last week on the pullback. See below section on $META.
Amazon (AMZN), GOOGL (Google), UBER, FOUR (Shift4), MELI, META, NU, NVDA, SOFI, MSFT (Microsoft)
Zane’s Portfolio
My Recent activity July/August/September 2026
-
I bought an initial position in Celestica (CLS) an AI data center equipment provider.
-
Nebius (NBIS) Bought some covered calls in Q1. As the price took off, I bought back my covered calls at a loss. My conviction on Q2 results became stronger and I wanted to maintain my full position. Q2 ER was a blow out and NBIS was up 30%, then down and then up. NBIS is now my largest holding but large daily price swings require some strong stomach.
-
Bought a small position SKHY (IPO Korean memory company SK Hynix) to complement my large Micron MU position
-
Sold in ELVA, battery company. ELVA is a solid company but I wanted more money into faster growing AI stocks.
-
Sold my small 1% bitcoin position for a small profit on buying the dip.
-
Sold in TARS, a pharma eyedrop company. TARS is a solid company but I wanted more money into faster growing AI stocks. TARS revenues were decelerating and no new product catalysts until later in 2027. Oops TARS jumped 20% right after I sold this dead dog.
-
Sold some NU Bank to buy faster growing AI stocks.
-
Sold half my AppLovin (APP) after a disappointing Q2 ER.
-
Sold most of Rocket Lab RKLB as profit on the SpaceX IPO halo effect. I hold a small position due to the risks with its first Neutron rocket launch.
-
Bought a start position in SIMO, a AI chip packaging company.
-
Sold covered calls against RBRK. Want to increase cash but don’t really want to sell RBRK.
-
Sold a little RBRK for some cash as it had gone up so fast the last 30 days and bought META.
-
Bought an initial position in META after settlement of law suits.
Cash holding park: (money market 1 year return SWVXX 4% and JAAA ETF 5.3%). Boring but safe.
Stock % Holding % Gain since Buy % Gain YTD
NBIS (Nebius) 17% 202% 172%
NVDA (Nvidia) 10% 216% 17%
RBRK (Rubrik) 6.50% 70% 16%
MU (Micron) 9% 171% 243%
ALAB (Astera) 8% 117% 71%
CRDO (Credo) 5% 32% 11%
FIGR (Figure Tech) 6% -7% -10%
RDDT (Reddit) 6% 18% -32%
PLTR (Palantir) 5% 27% -7%
SOFI (SOFI) 5% 82% -34%
CLS (Celestra) 3% 3% 10%
NU Bank 3% 22% -10%
APP (AppLovin) 2% 8% -53%
META 2% 5% -2%
SKHY (SK Hynix) 1.70% 12% 2%
HOOD (Robinhood) 20% 207% 0% Different account
NTSK (Netskope) 1% 9% -14%
Other stock holdings each <2%: SIMO, RKLB, FOUR, IONQ, GDX, JAAA, GDX
15% YTD gain is a disappointment this year. Recent pullbacks on RDDT and on AppLovin really hurt my total return to date. And the Iran War has hit my fintech stocks. Time will tell if they recover by year end. My current activity in September is to move to 10% cash. Why? In the last 90 days crude oil has jumped +20% and the 10-year Bond has jumped +20%. These are historically very radical moves in a short period against a strong American economy. I want some dry powder to buy stocks should the market pull back suddenly so I am taking some profits or writing covered calls. Below I have my many buy recommendations. I am holding on all of these purchases until there is some market clarity. If there is a strong dip, I’ll move on some of the below strong buy recommendations.
$APP AppLovin (advertising, mostly mobile)
APP faces challenges despite recent Q2 revenue growth +65%. The stock pulled back hard on Q2 ER based on rising concerns for sustainable growth. Q2 had a very small miss. Investors are questioning whether it can sustain >50% YoY revenue growth amid concerns for the weak advertising sector. On almost every conference call, an analyst asks why Meta or GOOGL can’t come in and out compete with Applovin and CEO Adam Foroughi says we have the better tech and better algorithms. Q2 missed FCF estimate by 29% and was blamed on some late quarter AI model corrections. APP stock price has been cut by over 50%. APP trades for 17x forward EPS. EPS is expected to grow by 67% this year and by 20% next year. Given the forecast, a hard bounce back will occur only if APP can deliver on Q3. So, it rides on how much you can trust CEO Foroughi. I’ll ride with him another quarter. Hold APP
$ALAB Astera Labs
ALAB has made a killing the last 2 years selling retimer chips to support GPU-GPU communications over high speed Serdes PCIe copper cables. But the AI rack communications technology has evolved very fast. ALAB seems to be keeping ahead of the technology and competitive threat. The main thing that should draw the investor’s attention is that Astera shifted its product portfolio from signal conditioners to becoming a much broader AI connectivity platform. Products related to the PCIe 6 generation accounted for more than 50% of the total sales, up from around one-third in Q1. But Scorpio X-Series just launched into volume production and should become the largest Astera product family in Q3, one quarter ahead of the company’s previous expectations. This shift is important as the switching business implies a substantially bigger content opportunity compared to Astera’s original business of retimers. The largest customer is hyperscaler Amazon. AMZN owns $40M of Astera shares, with a new warrant announced enabling them to own up to $466m of ALAB, based on Amazon buying $6.5B worth of future Astera products. ALAB has a risk of over concentration in one big customer.
2026 Q2 Revenue grew to $392M with growth of 104% YoY and EPS beating estimates to reach $0.80 beating by $0.11. The forecast for Q3 will be the main indicator that Astera is experiencing a totally new stage of growth. The revenue forecast will grow up to $540-$560 million, meaning 40% QoQ growth due to Astera Labs’ entry into an increasingly competitive switching market. This introduces materially higher competitive risk. Astera continues developing optical solutions, UALink, custom silicon, and future PCIe generations. However, the operating margins are still expanding because of the rapid revenue growth. Strong Buy ALAB.
$CLS Celestica (cloud equipment)
CLS delivered a strong Q2 CY26, with accelerating AI networking switch revenues and significant gross margin expansion. CLS is capitalizing on hyperscaler demand by investing $1B in capex, securing raw materials to support unprecedented growth and margin resilience. The recent $3B equity offering, while dilutive short-term (~8–10%), strategically funds capex and positions CLS for sustained high growth and margin expansion. Celestica trades at ~32x CY26 EPS that is now expected to grow 88%, while revenues are expected to grow 66%. The PEG ratio of 0.6 is a big giveaway implying how underpriced Celestica shares are compared to the 88% EPS growth this year. Celestica raised 2026 guidance to $20.5 billion in revenue and $11.30 in EPS, while Q3 growth approaches 69% YoY. Growth may be slowing but still aggressive. Revenue is a bit concentrated in 3 large customers. Component lead times exceeding 52 weeks are driving binding commitments and giving CLS demand visibility through 2028 and 2029. Strong Buy CLS.
$CRDO Credo (+31% YTD)) Credo’s Active Electrical Cables portfolio is its flagship product, featuring a new type of electrical cable with built-in signal recovery chips (PCI v6 SERDES). Such cables are important because they offer three advantages: they are thinner, lighter, and, most importantly, use almost 50% less energy. These cables are for high speed (800Gb) between AI GPUs and CPUs (<7 meters separation). CRDO is a competitor to ALAB but has a slightly different direction. ALAB is moving into switches and CRDO is moving into optical cables. So, I own both. Financials show hypergrowth: FY25 revenue was up 126% YoY to $437M, gross margins near 69%, and non-GAAP net margin above 50%. FY26 Credo’s AEC product ramp to remain intact at key customers including Amazon, Oracle Cloud Infrastructure, Meta, and Microsoft, driving over 60% expected revenue growth for Credo in fiscal year 2026-27. Strong Buy CRDO.
Figure Technology (FIGR) Figure’s is able to reduce HELOC (home mortgages) costs from $10,000 to $1,000, and has reduced process times from 42 days to 10. FIGR is basically taking loans and batching them together and then sell the security via a token, representing it on their own private ledger that they built. FIGR’s dominant market share, rapid growth, and capital-light model justify easily valuation.
2026 Q2 Revenue of $225.59M (+112.7% Y/Y) beats by $11.37M. Consumer Loan Marketplace volume was $4.3 billion in the quarter, a 132% increase from the prior year. Adjusted EBITDA increased 126% year-over-year to $119 million; Adjusted EBITDA margin reached 54.6%, an increase of 7 percentage points year-over-year.
FIGR has screaming growth by transacting HELOC loans faster and cheaper using blockchain. A forward P/E around 30x does not strike me as expensive if Figure can sustain anything close to this growth trajectory.
The long-term exciting thing about FIGR is its tokenization of assets, right now primarily HELOC loans. The total value of global financial assets — equities, fixed income, real estate, private credit, commodities, alternatives — sits north of $500 trillion. Today, almost none of it lives on programmable rails. So, the movement of assets electronically today is slow like a horse and buggy. FIGR has the blockchain to move any kind of asset. I think FIGR may have gotten caught up and associated with the crash in crypto and Bitcoin names more recently. However, there business is pretty well insulated from swings in crypto. This major risk to FIGR is its current dependency on stable home loan interest rates. The current macro housing economy is wishy washy. Last year the Genius Act was passed and it allows banks to hold stable coins and to put them on the blockchain rail. Expect banks to offer 24 hour loans against a customer’s different wallet that have been tokenized. The wild card now is the pending next crypto Clarity Act in Congress. Approval could pump up FIGR. Hold FIGR.
$MU Micron (memory chips)
Micron fell in tandem with other AI stocks in July and is now recovering. I am still of the opinion that memory is no longer cyclical for at least the foreseeable future. In July, Samsung and SK Hynix announced $2T in new capacity to address the scarcity. This would usually be the next leg in the cycle where supply arrives and prices collapse. Both SK Hynix’s and Samsung’s dismissed oversupply concerns and explained their capacity expansion is “based on demand visibility” and warned that shortages may persist through 2028.
Micron was the first company to structure their deals in long-term contracts and commitments. Now the rest of the industry is following the same roadmap. Micron is building massive manufacturing facilities in Idaho and Virginia. The CEO says memory requests exceed his ability to fulfill by at least 50%.
Even at the second biggest in my portfolio, I still want to hold my Micron position. Just like Nebius, the AI buildout is here, and I want a chunk of it. Hold MU.
$NBIS Nebius (neocloud)
Nebius is a neocloud provider of AI data center services with an international footprint. It provides a full hardware and software AI stack for customers. Hyperscalars (META and MSFT) now make up the majority of revenue using just hardware but with a lower gross margin. The primary risk with NBIS is execution while maintaining good margins and therefore profit. The second risk is delivery on committed data center contracts. Q2 earnings last week were incredible, driving the stock even higher.
The next step of the thesis depends not on growth numbers but on the generation of revenue out of gigawatts without excessive stock dilution The problem of Nebius still is the massive amount of capital required to build the business. Q2 CapEx of the company stood at $5.7B. Also, the company issued 12.7 million shares through its ATM program with an average price of $224, bringing approximately $2.8 billion. Recently, the company underwent another convertible issuance, which caused the correction from almost $300 to $200. This is where the thesis becomes much stronger. If customers fund the infrastructure construction before deployment, contracted cash flows provide cheap asset financing, and operating cash flows scale with revenue, then equity capital is not required anymore.
2026 Q2 Revenue surged 514% YoY to $575 million, while ARR climbed 56% sequentially to $3 billion. Customer prepayments cover 50% to 70% of project CapEx, supporting a massive $20 billion to $25 billion 2026 expansion. Payoff of capex is now 1 yr 10 months down from 2-3 years
Margins in Nebius AI reached 50%, while the power contract target was raised to 5 GW without compromising the guidance on ARR of $7-$9 billion. 50% adjusted EBITDA margin, while group EBITDA was $236 million and the margin was 41%. High depreciation undermines the headline 50% adjusted EBITDA margin.
CEO Arkady says he is holding a small portion of capacity for short term, high margin contracts (e.g. 6 months): 30-50% margins. Demand is not the bottleneck, and the ability to balance mid-term contracts, premium short-term capacity, and long-term agreements gives Nebius valuable pricing and financing flexibility. I have belief in CEO Arkady, a tough, cloud experienced, old Russian survivor. He talks straight and tough with high exposure. And I like the NBIS AI diversification into Europe which badly wants AI services. Arkady knows how to navigate these regulators. My only neocloud hold is Nebius and is my largest position.
For at least 2027, I do not see this AI supply/demand imbalance changing. Long term will the neoclouds provide a software stack that will generate sufficient revenue and margin? dunno but that is a long time and right now all boats will float. So, for now, I will float in my boat with one eye open and laser focused on the NBIS margins. Buy NBIS on pullbacks.
$NVDA NVDIA (AI chips and servers)
NVDA is a safe way to continue to play the AI market. In fact, it is almost a defensive stock in relation to the other AI chip companies. Its growth this year has been down right pedestrian. Right now, NVDA controls >50% of the AI chips and software stack. The NVDIA GPU AI chips are the best in the world. And its AI stack is the best general-purpose suite in the world. Even given strong competition, I still believe NVDA has stable growth ahead in 2027 and beyond. It is estimated that we will have General Artificial Intelligence in 2-3 years, meaning human reasoning qualities. AI-based girlfriends and boyfriends online are here and love is in the air. Competition is building with Broadcom XPUs and AMD CPUs for inference engines. But NVidia’s revenue growth rate and margins have stayed high: Q1 GAAP margin was 74.9% and free-cash-flow was $45.5 billion, up 16.7% sequentially. Yet NVIDIA currently trades with a forward P/E that is significantly less than Apples despite growing earnings much faster and having a stronger balance sheet. I hold NVDIA as my foundation stock. At $216, Nvidia trades around 23x forward GAAP earnings despite consensus forecasting fiscal 2027 revenue growth of approximately 83%. Nvidia is becoming embedded in AI financing through investments, take-or-pay commitments, and more than $500 billion of external capital.
At roughly 15x FY2028 earnings, valuation appears disconnected from 66% expected EPS growth and continuing estimate upside. Strong Buy NVDA (up only 15% YTD)
NTSK NetSkope (-42% YTD) My old boss just took his company IPO last year so the public track record is short and the company is currently unprofitable. NTSK is growing +28% YTD and is in the cloud traffic brokerage business. It should see a great benefit from burgeoning agentic AI transactions and increased cyber security demands. IMO this cyber security company will benefit from the high growth of AI Agentic traffic. At 6x sales it is more affordable that Palo Alto Networks (PNET) at 18x sales and growing similarly. If NTSK can maintain growth and margins, profits and repricing will follow. Buy under $14.
$PLTR Palantir (AI services commercial and government)
I picked up PLTR on a recent crash of about 35%. I never liked the price and still do not. And many of their government contracts have no transparency. But its incredible performance made to bite at under $130/share. Hearing the CEO describe the moat as providing secure AI services while protecting the data and resulting tokens as intellectual property. Specifically, cloud AI services cannot get access to your company’s tokens to use in their inference engines. I can see the commercial advantage. PLTR is up 40% since my purchase. The stock trades at a market cap of $416 billion, while generating just $6.2 billion in trailing revenue. That is a price-to-sales ratio of 73, making Palantir one of the most richly valued stocks out there today.
Palantir is also making its AI pitch a lot more concrete. Its expanded NVDA partnership, as reported by Barron’s, puts its robust software to work within the chipmaker’s walled-garden-like supply chain. At the same time, a new Nebius agreement offers its customers greater control over computing infrastructure and models.
Palantir delivered 93% revenue growth while U.S. commercial revenue surged 149%, demonstrating exceptional monetization beyond customer acquisition. Adjusted operating margin reached 62% and free cash flow margin hit 63%, despite adjusted expenses increasing 37% year-over-year. RPO doubled 103% to $4.9 billion, with $2.81 billion longer-term, suggesting customers are becoming increasingly embedded within Palantir.
I love today’s announcement that PLTR will partner with NBIS. The underlying thesis is centered around enterprise data. Arkady’s view is that companies will increasingly want to run open-weight models on their own proprietary data, repeatedly improve those models inside their own domain, and retain control over the resulting data and intelligence rather than continuously feeding it back into external commercial models. Palantir already has the enterprise software layer and the tooling required to orchestrate those workflows. What it needed was the infrastructure underneath it. That’s where Nebius comes in.
NBIS CEO Arkady “We provide the whole stack up to our Token Factory. They take it, add their tools on top of it, and their customer base. For us, it is an excellent channel to enterprises.” This potentially gives Nebius access to an enterprise customer base that would have taken considerably longer to penetrate organically. Hold PLTR (or sell covered calls)
$RDDT
Reddit Q2 Revenue came in at $802.9M (61% YoY) beating guidance by 11.8%, the eighth (!) consecutive quarter of 60%+ growth. They guided Q3 to $865M (48% YoY) and with a similar percentage beat would be just around 60% growth again next quarter.
The main issue people had with earnings and the reason the stock declined after was in their US (Daily Active Users per quarter) DAUq metric. It actually fell from 53.5M to 53.2M (0.6% QoQ). Global DAUq was 130.3M (up 18%) and WAUq was 514.6M (up 24%). So, they are still growing globally and some of those daily users who dropped off are still checking weekly. Reddit is successfully exercising their ability to gain revenue per user (ARPU), but having more trouble growing that user base. It means revenue monetization can certainly continue, but a stagnant US DAUq figure will lead to the party ending sooner rather than later.
Additionally, there was a WSJ report earlier in July that Reddit had internally discussed ending Google’s ability to use its content for AI training as the $60M per year deal nears expiration. The stock dropped on this story but I view this as a positive. I think it speaks to the strength of Reddit’s hand when it comes to their positioning and ownership of the training data. They wouldn’t hold out like this unless they think they can negotiate better terms. Open AI is also reporting fewer RDDT citations. I believe that RDDT can leverage their data in exchange for much more money. If not, I will sell.
Overall, the business quality and metrics continued on their stellar trajectory. I would normally want to hold tight, no buys or sells until there is more clarity in this situation. But RDDT price action has gone south. RDDT’s 61% YoY revenue growth, rising ARPU, and minimal CapEx position it for scalable, high-margin expansion as human-generated content gains value in the AI era. Any announcement on an agreement with GOOGL or Open AI will send RDDT stock flying. But this is a big gamble so I will not add to my position. Hold RDDT
RKLB Rocket Lab (+78% YTD) I hate myself for selling this stock at $24 and again at $55. But I bought shares at $7 in early 2025 for grins. I am a space junky and I saw a huge revenue growth but no earnings, huge debt, and a lot of rocket launch risk. Owning RKLB would just be for fun. Since then, RKLB has collected billions of dollars of contracts from the US government. I saw it popping on a halo effect from the coming SpaceX IPO last June. The stock rocketed to $124 then back to $70. So, I bought low and sold high and took some special situation profits. A small Watch RKLB
$RBRK Rubrik This is a cyber security company growing revenues +40% YOY. It has a niched in data and system recovery after a cyber breech. Key metrics include 36% subscription ARR growth, 120%+ net retention, and an increase in large enterprise wins, validating its cyber resilience strategy. The best way to value this company is through revenue-based valuation metrics because RBRK still lacks profitability. Customers with $100,000 or greater in subscription ARR increased 27% year-over-year to 2,505 customers. These larger customers increased from 82% of the total subscription ARR last year to 85% this year, indicating Rubrik is becoming a strategic partner to large enterprises, which aligns with the high-stakes, board-level issue of cyber resilience. I would rather own RBRK at 12x sales that Crowdstrike CRWD at 35x sales or PANW at 18x and growing slower 26% YoY. RBRK is no longer a strong buy with the recent runup to $100 after Q2 ER. Buy under $80.
#############################################################