G.L.A. INVE$T RE$EARCH · Infinite Innovation

Mag 7 Update & Outlook · Wednesday, Oct. 7, 2026

Week of Sept. 30 – Oct. 7 · DRAFT for owner review · Prices as of the Oct. 7 close (ET)

Happy Thursday to the Greatest Living American’s!

This is a brand new series I decided to write, The Mag 7 Newsletter which will keep you updated on the 7 companies that have dominated the last 15 years. The important thing to understand is that the folks who traded these names made a fraction of what the buy and holders made. Imho the best strategy with the stock market is to build positions in 7-10 tech juggernaughts like the mag 7 or “FAANG.

The Big Lesson: Buy, Hold, and Let the Innovators Compound

Every issue starts with the one chart that explains why G.L.A. exists.

If you had put $10,000 into an equal-weight basket of the five Mag 7 companies that were public 20 years ago (Apple, Microsoft, Alphabet, Amazon and NVIDIA) on Oct. 6, 2006, and then did nothing (no trading, no rebalancing, no timing), you would have about $1.75 million today. That’s a 29.5% average annual return. The same $10,000 in the S&P 500 (SPY) grew to about $83,000 (11.2% a year), and the Nasdaq-100 (QQQ) to about $214,000 (16.6% a year). The FAANG names that were public then (Apple, Amazon, Netflix, Google) turned $10,000 into about $1.37 million (27.9% a year). (G.L.A. calculation from Yahoo Finance adjusted closes, dividends reinvested.)

Now compare that to the “smart money.” In 2007, Warren Buffett bet $1 million that a plain S&P 500 index fund would beat a hedge-fund industry pro’s picks over 10 years. Protégé Partners chose five funds-of-funds holding more than 200 hedge funds. Buffett’s final scorecard: the five funds averaged 0.3% to 6.5% a year, while the index fund made 8.5% a year (+125.8%). In Buffett’s words: “Performance comes, performance goes. Fees never falter.”

We added the innovators to Buffett’s scorecard for the exact same 2008–2017 window:

The Mag 7 names that were public then averaged 20.2% a year (+531%). The FAANG names averaged 33.7% a year (+1,720%). That is about three times the best of the five hedge-fund portfolios (6.5% a year), with no 2% fee and no 20% cut of the gains. (Our SPY calculation for this window came to +125.0%, within a hair of Buffett’s +125.8%, a good check on our method.)

And it isn’t just ancient history. In 2025, hedge funds had their best year since 2009: the HFRI Fund Weighted Composite rose 12.6% . An equal-weight Mag 7 basket returned 22.7%, the S&P 500 (SPY) 17.7%, and the Nasdaq-100 (QQQ) 20.8%.

The honest part (because we teach, we don’t sell): a 20-year chart picked today has hindsight built in. Nobody called these the “Mag 7” in 2006. So here’s a test with no hindsight: what happened after Wall Street gave each basket its name? FANG was coined by Jim Cramer in 2013, Apple was added to make FAANG in 2017, and BofA’s Michael Hartnett coined “Magnificent Seven” in May 2023 . Buying each basket the day it became famous, and just holding, still beat both indexes:

Total return, dividends reinvested, equal-weight, no rebalancing. G.L.A. calculation from Yahoo Finance data through Oct. 7, 2026.

Two more honest notes. First, NVIDIA does a lot of the heavy lifting: take it out and the 20-year basket drops from 29.5% to 25.4% a year. That’s still more than double the S&P 500. Second, “beat every hedge fund” is a claim nobody can verify fund by fund. The top 10% of hedge funds made 62.7% in 2025 (HFR, same source). What the evidence does show is that a simple buy-and-hold basket of these innovators has beaten the hedge-fund industry average and the S&P 500, Nasdaq Composite and Nasdaq-100 over every long window we tested.

G.L.A. VIEW: This is the Investing Constitution in one picture. Our approach is to concentrate in 6–10 world-changing companies, buy and hold for 5–10 years, add on weakness with dollar-cost averaging, never use margin, and let compounding do the work. The weighing machine rewards patience, not activity.

The Week in Brief

The Mag 7 led again. From last Wednesday’s close (Sept. 30) to today’s close, an equal-weight Mag 7 basket rose +2.9%, versus +2.0% for the S&P 500 and +2.5% for the Nasdaq. Six of seven names rose; only Meta slipped.

(yesterday and today) Today was a pause. The S&P 500 fell 0.22% to 7,801.77, ending a four-day win streak, and the Nasdaq also fell 0.22% to 27,538.69. The 10-year Treasury yield touched 5.365%, its highest since April 2002, before a strong auction calmed bonds. Minutes from the Fed’s September meeting showed most officials saw another hike as likely by year-end, though markets priced under 20% odds of an October move (Invezz; WSJ).

Year to date, the Mag 7 is a tale of two groups. NVIDIA (+27.3%) and Apple (+23.8%) are well ahead of the S&P 500 (+14.0%). Tesla (−16.0%) is the laggard, and that drags the equal-weight basket to +11.2%. Over the last 12 months, the Mag 7 average (+15.6%) has roughly tied the S&P 500 (+16.2%), but leadership has been rebuilding since July:

G.L.A. VIEW: A year of “sideways” for the leaders is not a broken thesis. It’s how the weighing machine resets prices before the next leg. That’s when patient investors add, not panic.

The MAG 7- Company by company

Prices are Oct. 7 closes. “Week” is Sept. 30 to Oct. 7 close. Source for prices: Yahoo Finance.

NVIDIA (NVDA): $237.47 · Week +4.0% · YTD +27.3% · 0.7% below its 52-week high

What mattered: NVIDIA spent the week chasing a $6 trillion market value, a level no company has reached. Bloomberg credits two things: a strong revenue outlook and the extra $150 billion buyback, which Jensen Huang said “reflects our confidence in the long-term opportunity ahead”. Its last quarter (Q2 fiscal 2027) brought in $96.2 billion of revenue, up 106%, with data center at $89.0 billion, and it guided to about $108 billion next quarter, assuming no China data-center sales. NVIDIA also showed up inside Microsoft’s new Surface Laptop Ultra, which runs on NVIDIA’s RTX Spark chip (see Microsoft below).

Why it matters for long-term innovators: NVIDIA is the toll road of the AI platform, and now it’s moving from the data center onto the PC. Outlook (G.L.A. VIEW): The next test is whether growth stays supply-limited, not demand-limited. Watch the hyperscalers’ capex updates in late October.

Microsoft (MSFT): $529.76 · Week +3.3% · YTD +9.5% · 2.3% below its 52-week high

What mattered: At a San Francisco event today, Microsoft laid out “hybrid intelligence” for Windows: AI agents that run locally when it makes sense and reach the cloud when they need to. Microsoft Execution Containers, a security layer that limits which files and networks an agent can touch, is now generally available, and Copilot will be able to take actions on your local files with permission. Pre-orders opened for the Surface Laptop Ultra (from $2,599, ships Oct. 16) and the $5,999 Surface RTX Spark Dev Box (ships in November). Engadget noted Microsoft is pushing harder on agents after Meta’s Muse found a big audience fast .

Why it matters: Whoever makes AI agents safe for businesses wins the enterprise, and enterprise is Microsoft’s home turf.

Outlook (G.L.A. VIEW): Agents that do real work, billed by usage, could let Microsoft grow revenue with the work instead of with headcount.

Amazon (AMZN): $259.92 · Week +4.3% · YTD +12.6% · 8.5% below its 52-week high

What mattered: Amazon rose 1.4% today on bullish analyst calls. Tigress Financial raised its target to $385 (Oct. 6), TD Cowen reiterated $350, and Rosenblatt lifted its target to $360. TD Cowen says Amazon’s AI business exited Q2 at roughly a $25 billion annual run rate, and Amazon expects AWS to stay capacity-constrained through 2026 and 2027 even as it plans to double capacity by the end of 2027. AWS’s Matt Wood said the cost of a given level of AI (outside frontier models) is falling by one or two orders of magnitude every three to six months .

Why it matters: That cost-decline line is the heart of our thesis and ARK’s: when intelligence gets cheaper, demand explodes.

Outlook (G.L.A. VIEW): Free cash flow will look tight while Amazon builds, so judge it on AWS growth and backlog, not quarterly cash flow. Analyst targets are opinions, not facts.

Alphabet (GOOGL): $350.50 · Week +1.9% · YTD +12.0% · 12.9% below its 52-week high

What mattered: Finland’s permitting regulator (LVV) ordered Google’s subsidiary to stop environment-altering prep work at two planned data-center sites, Muhos and Kajaani, by Oct. 23, until environmental impact assessments are done. The sites are part of Google’s roughly $15 billion Finland plan, its largest investment in Europe. Google said it had “fallen short of our own high standards in this instance” and will comply; its written response is due Oct. 14 . The stock shrugged it off, up 0.8% today.

Why it matters: It’s a delay, not a derailment, and a reminder that power, land and permits are now the bottlenecks of the AI build-out. Per ARK Invest, Google Cloud growth accelerated to 82% year over year in Q2 .

Outlook (G.L.A. VIEW): Alphabet owns its chips, models and cloud, and after Tesla it’s the furthest Mag 7 name from its 52-week high. Earnings are Oct. 28.

Meta Platforms (META): $721.31 · Week −0.5% · YTD +9.3% · 7.2% below its 52-week high

What mattered: Meta was the week’s only Mag 7 decliner and fell 2.4% today; no single cause was confirmed. On Oct. 6, Meta and Sierra (Bret Taylor’s company) announced the Personal Agent Protocol, an open standard for how personal AI agents like Muse sign in, act for users, and buy things from businesses. Partners include Shopify, Stripe and Walmart, and a first version is due later this month . Citigroup estimated Muse could generate more than $27 billion a year by 2030, about $23 billion of it from transactions . Muse is still the top free iPhone app, but CNBC asks the right question: can it become a habit? As IDC’s analyst put it, “Meta’s ad business can pay for a lot of waiting” .

Why it matters: Agentic commerce could be Meta’s first big revenue line beyond ads, and open standards are how agents get let into stores like Walmart. Risks we’re tracking: reported security fixes before Muse’s launch (404 Media, via SiliconANGLE) and heavy capex.

Outlook (G.L.A. VIEW): Citi’s number is a forecast, not guidance. We want to see usage and paid plans grow.

Tesla (TSLA): $377.81 · Week +6.5% (best in the Mag 7) · YTD −16.0% · 22.9% below its 52-week high

What mattered: Tesla carried last Friday’s Q3 delivery beat (486,532 vehicles, plus 13.7 GWh of energy storage) into the week’s best gain . In Europe, the EU vehicle committee discussed supervised FSD again on Oct. 6 but did not vote; a vote is now December at the earliest. Germany backs approval, while Sweden and France have raised concerns .

Why it matters: Tesla touches three of our five platforms: Energy Storage, Robotics (Optimus, Cybercab) and AI (FSD).

Outlook (G.L.A. VIEW): Wall Street still prices Tesla like a carmaker, which is why it’s the Mag 7’s biggest YTD laggard and the most volatile name here. Volatility is vitality. Q3 earnings are Oct. 21, the next big test of the autonomy story.

Apple (AAPL): $336.67 · Week +1.1% · YTD +23.8% · 1.3% below its 52-week high

What mattered: Bloomberg’s Mark Gurman reports Apple will introduce a smart-home hub (a 6-inch display), a new HomePod mini and Apple TV on Oct. 13, and Bloomberg and TechCrunch report a lock, thermostat and doorbell lineup co-developed with LG (MacRumors; TechCrunch). The iPhone Duo foldable opens for pre-order Oct. 16, but a supply-chain source told China’s Jiemian News that Foxconn’s assembly yields were only slightly above 60% in mid-September, so early supply may be tight (same MacRumors report; unconfirmed by Apple). Gurman also reports CEO John Ternus wants faster, more frequent product launches.

Why it matters: Apple is the distribution layer for Intelligent Devices, the part of the AI platform that reaches your pocket and your home.

Outlook (G.L.A. VIEW): Apple has quietly been the Mag 7’s second-best stock this year. Fiscal Q4 results arrive Nov. 2; the Duo won’t show up until the December quarter.

The Innovation Thesis: Who Pays for the AI Build-Out?

From ARK Invest (Big Ideas 2026): ARK estimates hyperscalers will spend more than $500 billion on capex in 2026, nearly 3x the $135 billion of 2021. Investment in data-center systems was about $500 billion in 2025 and could roughly triple to ~$1.4 trillion by 2030. ARK also notes that while tech capex as a share of GDP is at levels not seen since 1998, tech valuations are “a fraction of” their dot-com peak. ARK estimates inference costs have dropped more than 99% in the past year by some measures.

G.L.A. VIEW: The Mag 7 are the ones writing these checks, and the ones collecting the tolls. Cost declines are the engine of all five innovation platforms. The risk is timing: spending comes before earning, so cash flow can look ugly before it looks beautiful.

Outlook: What We’re Watching (G.L.A. VIEW, not trading advice)

1. Apple’s smart-home launch (Oct. 13) and iPhone Duo pre-orders (Oct. 16), including any sign of supply limits.

2. Google’s response to Finland (due Oct. 14) and whether other European permits slow down.

3. Tesla Q3 earnings (Oct. 21): margins, energy storage, and Cybercab and Optimus timelines.

4. Alphabet earnings (Oct. 28) and the late-October hyperscaler reports. Microsoft, Meta and Amazon dates are expected in late October; we’ll confirm them next week.

5. Rates. A 10-year yield above 5.3% is a headwind for every growth stock. We don’t time macro, but we watch it.

The Takeaway

Twenty years of data say the same thing Buffett’s bet said, only louder:

• owning great businesses and sitting still beats paying professionals to trade. The innovators we cover beat the hedge-fund average, the S&P 500 and the Nasdaq, even when you buy them only after everyone knows their name. That’s not luck. It’s what happens when companies ride AI, robotics and energy storage cost curves for decades.

• Our approach is simple: know the businesses, concentrate in the highest-conviction names, add on weakness, never use margin, and give the weighing machine time. Weeks like this one, with yields at 24-year highs and one name down while others rise, are noise. Zoom out.

— G.L.A. INVE$T RE$EARCH · Infinite Innovation

How we calculated: Long-term returns are total returns (Yahoo Finance adjusted closes, dividends reinvested), equal-weight at the start date and never rebalanced. Weekly, year-to-date and 12-month figures are price-only closes pulled about 5 PM ET on Oct. 7, 2026. Hedge-fund figures come only from the sources cited. Charts and data files are available on request.