BTC_Chopsticks

99% of Lam Research's $1.50B quarterly revenue increase came from systems sales.

$LRCX added $1.48B in systems revenue versus June, while customer support-related revenue and other grew by just $14M.

That makes this quarter's growth highly dependent on new equipment spending. The investor question is how well earnings hold up when chipmakers slow their next round of fab investment.

The counterpoint: support-related revenue is still growing about 40% year over year, and new equipment expands the future installed base.

My test: can that installed base support profit growth through the next equipment-spending slowdown?

Tesla's free cash flow went from -$1.09B to +$1.53B in one quarter. That's a $2.62B rebound, but the bridge matters more than the headline.

$1.19B came from changes in net operating assets and liabilities, $1.06B from lower capex, and $0.38B from net income plus noncash adjustments.

These are real cash-flow movements, not additional profit.

My test: can Tesla sustain positive cash generation after capex as working-capital contributions fluctuate?

That would strengthen its ability to fund expansion internally, without yet proving attractive returns on AI investments.

Half-price model tokens can mean only a 25% smaller AI bill.

Take a hypothetical DeepSeek Flash batch: 1M cache-miss input tokens, 1M output tokens, 1,000 successful Search requests and 1,000 successful Fetch requests.

Using listed rates after November 1's integrated-tool billing transition:

Peak: $1.50 tokens + $1.50 tools = $3.00.
Off-peak: $0.75 tokens + $1.50 tools = $2.25.

Tokens cost 50% less. The combined bill falls 25%.

For AI application investors, compare cost per accepted result after tools and rework. This is hypothetical pricing arithmetic, not actual customer usage, completion quality or profitability.

Free $USDC redemption isn't the same as cash available now.

Circle Mint's new structure takes effect on or after October 19, with existing customers moving at their next billing period. On a $1M gross redemption, Standard has no Circle fee and processing up to two business days; Express is $500 before eligible credits. Both still require bank settlement.

Mint is institutional, not a universal retail off-ramp. CPN activity is exempt from Express fees, and separate agreements can change terms.

For investors assessing stablecoin liquidity, compare time to usable cash with the net fee. Faster processing is worth paying for only when the liquidity benefit exceeds its cost.

State Street pulled $69B into ETFs in Q3, but net inflows across investment management were $7B.

Other commingled funds lost $65B; separately managed accounts added $3B.

That isn't $69B of new money across the franchise, nor proof the same investors switched funds.

The shareholder opportunity is retaining assets that earn attractive fees. Management fees already rose 20% year over year, helped by markets and long-term inflows.

I'd compare fee growth with retention and costs. ETF wins can be valuable without translating dollar-for-dollar into revenue or profit.

Stablecoin adoption has a competitor hiding in plain sight: the interest paid on bank deposits.

An October 13 BIS working paper, using US data from January 2019 to June 2025, finds that higher deposit rates predict slower stablecoin growth. The relationship is stronger for USDC than USDT.

The opportunity for stablecoin businesses depends partly on how attractive bank cash looks, not just how many wallets they reach.

That doesn't prove stablecoins are draining the banking system or tell us what today's token prices should do.

For investors, compare funded balances and repeat payment use with deposit-rate competition before extrapolating adoption.

Interactive Brokers is more than a bet on trading volume.

In Q3, net interest income was $1.154B, versus $602M in commissions. Interest supplied 60.3% of net revenues, up from $967M a year earlier.

More client balances and borrowing can expand earnings alongside trading. But that also makes rates and funding costs part of the investment case.

A growing account count isn't enough if revenue per balance compresses. Compare net interest income with average client balances, not just trade counts.

This is a revenue mix, not a profit split or proof the stock is cheap.

Bitcoin and Bitcoin-treasury shares carry different forced-selling risks.

MSCI has finalized a non-operating-company screen. The final affected-constituent list arrives November 11, with changes effective at the November 30 close.

That is not a confirmed deletion list today.

For investors choosing a treasury stock over BTC, index eligibility adds a risk beyond the coin's price. Index funds selling shares would not itself require the company to sell Bitcoin.

Check the final list and the company's financing needs separately. Don't turn possible equity selling into a guaranteed BTC liquidation.

One positive ETF day did not reverse the three-session total.

U.S. spot Bitcoin ETFs took in $175.7M on October 13. Across October 12-14, they still lost a net $151.1M. Ether ETFs lost $190.1M over those same completed U.S. sessions, according to Farside.

If I'm looking for returning demand, I want persistence, not one green row.

This is a short window and only the ETF channel. It doesn't prove where prices go next, or which asset faces more pressure relative to its fund size.

TSMC spent NT$443B on capex and still generated NT$477B in free cash flow in Q3.

$TSM 's October 15 results show NT$919.77B of operating cash flow and NT$476.56B after capex, versus NT$139.38B a year earlier.

That gives the company room to fund growth. It doesn't prove the stock is cheap.

This cash measure is before dividends. Working capital and spending timing can move it between quarters.

Repeat cash generation as new factories ramp would strengthen the investment case. Spending that grows without adequate returns would weaken it.

🏡 回到家才发现，沙发上最舒服的位置已经被 TRON Bull 霸占了。😂

夕阳刚好照进客厅，这小家伙坐在那里，感觉比我还懂得享受生活。

原本只是参加 TRON 社区活动带回来的小礼物，没想到现在已经成了家里的一员。

平时不是盯行情，就是忙着研究各种 Crypto 热点。偶尔放下电脑，喝杯咖啡，看看窗外的夕阳，也挺不错的。☕🌅

不过照这样下去，TRON Bull 是不是也该交点房租了？😂❤️

@justinsuntron @TRON DAO @TronDao_THA
#TRONGlobalFriends #TGF

ASML can sell more chipmaking capacity without selling another new machine.

In Q3, $ASML 's service and field-option sales reached €3.289B, up 19.1% from Q2. Management says demand for productivity upgrades helped lift company-wide gross margin to 57.5%.

The opportunity: earn more from equipment customers already own, alongside new-system sales.

But upgrades aren't a guaranteed subscription, and customer expansion plans aren't orders.

I'd watch repeat service-and-upgrade demand and cash generation through the cycle. Slowing customer utilization or weaker upgrade demand would weaken the thesis.

JPMorgan earned $15.3B. Net buybacks were $664M. The opportunity is what retained capital earns next.

$JPM Q3 2026 release says management reduced repurchases to prioritize investment and clients.

That benefits shareholders if reinvestment beats buying back shares at the available price.

I'd test earnings per share against credit losses and expenses. If growth requires worse underwriting or rising costs absorb the gains, retaining capital hasn't proved its value.

Fewer buybacks can be sensible. They still need to earn their place.

AI agents could make CPUs a bigger part of AMD's AI business.

On October 12, $AMD introduced Helion, an open rack blueprint with a CPU-only configuration supporting up to 32,768 cores. Configurations are planned for November 2026.

Agents need compute to execute code and run workflows, alongside the GPUs handling model inference. That could create another route to EPYC sales.

But a blueprint isn't an order, and some spending may replace existing servers.

My decision: WATCH.

I want paid deployments and repeat orders that improve earnings and cash generation. Delays, weak adoption or substitution without additional profit would weaken the case.

AI could shrink the coding bill and create a new revenue stream for EPAM.

On October 5, $EPAM announced specialized training-data, model-evaluation and reinforcement-learning environment services.

The opportunity: charge AI developers for realistic enterprise tasks and reliable ways to test whether agents complete them.

Its engineering experience could help. But the announcement doesn't establish paid adoption or material earnings from this offering.

My decision: WATCH.

I want paid engagements, repeat business and stronger cash conversion, not just more AI partnerships.

The counterargument matters: automation could compress traditional service revenue faster than this business grows.

Shareholders benefit if EPAM earns more from teaching and testing AI than it loses from work AI replaces.

TSMC's revenue base is nearly 51% larger than a year ago. September's small monthly dip doesn't erase that.

Adding
$TSM's reported July, August and September sales gives approximately NT$1.494T for Q3, versus NT$990B last year.

That's 50.9% growth, calculated from its unaudited monthly reports.

September alone fell 0.6% from August.

For shareholders, the opportunity is whether that larger revenue base converts into stronger earnings and cash after investment.

October 15 earnings are the next test: margins, capital spending and the outlook.

My decision: WATCH.

Revenue growth alone doesn't establish a cheap stock. Margin erosion, weaker guidance or poor cash conversion would weaken the case.

Compare the whole quarter, then check what shareholders keep.

AI chips get the attention. Broadcom can also get paid to connect them.

$AVGO’s Tomahawk 6 delivers 102.4 Tb/s of switching capacity, twice Tomahawk 5’s 51.2 Tb/s. It is already shipping in production volume.

Its October 8 OCP preview puts switches, network interfaces and integrated optics in focus.

The investment opportunity: larger AI clusters need to move data between accelerators. Broadcom sells part of that connectivity.

There is already a substantial business behind the thesis: Q3 AI semiconductor revenue reached $16.7B. But that includes custom accelerators and networking, not networking alone.

Twice the bandwidth does not mean twice the revenue. More capable chips can reduce the hardware needed for a given workload.

My decision: WATCH.

I want customer deployments to translate into profitable shipments and durable cash flow. Slower adoption, pricing pressure or weaker customer spending would weaken the case.

Connecting more compute can be valuable. The stock still needs a price that leaves room for shareholder returns.