Daily Tech Report

Research cutoff: Friday, September 4, 2026, roughly 08:30 BST.
U.S. price cutoff: Thursday, September 3 regular-session close, plus confirmed after-hours information where explicitly noted.
Horizon: roughly the next 1–7 U.S. trading sessions.
Pre-market note: It is only about 03:30 ET at this cutoff, before the most liquid U.S. single-stock premarket. Reuters' latest global-market snapshot has Wall Street futures practically flat. An earlier Asia snapshot had Nasdaq futures around +0.3% and S&P 500 futures around +0.1%, so the mild overnight bid has faded toward neutral. I am therefore using Thursday's confirmed closes and today's macro setup rather than pretending thin early quotes are meaningful.

The quick read

Friday is mostly about one thing: the U.S. jobs report at 8:30 a.m. ET / 13:30 BST.

Thursday gave tech a very strong setup. The S&P 500 rose 1.06%, the Nasdaq gained 1.40%, and the Dow climbed 1.18% after Fed Governor Christopher Waller pushed back against the rush to price another rate hike. The market-implied probability of a September hike dropped to roughly 50%, from above 63% a day earlier.

But there is a catch. Brent crude is still around $95.5 a barrel, the U.S. 10-year Treasury yield is around 4.76%, and Thursday's services data showed input prices at a three-year high. So the inflation problem has not disappeared; the market simply got some relief from the idea that the Fed does not need to react immediately.

Today's payroll consensus is only +56,000 jobs, after July's -23,000, with unemployment expected to stay around 4.1% and annual wage growth to slow to roughly 3.0%. The forecast range is unusually wide, and some economists think unemployment could tick to 4.2%.

The important nuance is that Waller suggested this payroll report may not decide the September 15–16 Fed meeting unless it is a major surprise. Next week's CPI may matter even more. That makes the market reaction less binary than usual: a modest payroll number can still be interpreted as consistent with a slow-hire, slow-fire economy rather than an imminent recession.

Inside tech, Thursday was anything but uniform:

• Tesla +5.42% and Meta +3.01% were the strongest names in this group.
• Microsoft +2.68% confirmed the catch-up move after its new Azure disclosure.
• Nvidia +1.80% rose after announcing a $12.93 billion acquisition of Hugging Face, a major strategic move into the open-model/developer ecosystem.
• Google +1.59%, Amazon +1.54% and Apple +1.00% all participated in the rally.
• AMD fell 0.20% despite the Nasdaq's 1.4% gain.
• Broadcom fell 2.74% as investors continued to punish the slightly soft near-term revenue guide despite its huge 2027–2028 AI forecast.

My working view this morning:

The broad tape improved, but today's best information is the divergence inside tech. META/MSFT/NVDA are showing real price confirmation. AMD and AVGO are not. The jobs report can amplify either side, so I would make the first 30–60 minutes after 13:30 BST prove the direction instead of guessing it before the data.

Market pulse before the U.S. open

The most important thing here is the starting point. Thursday's rally came from a rates repricing, not from oil suddenly collapsing or inflation risk vanishing. If payrolls and wages surprise materially hot, the 10-year can move back toward recent highs and hit the highest-beta tech names quickly.

The tech board

The price zones below are observation/decision areas derived from recent daily highs, lows and closes. They are not analyst targets, guarantees or automatic entries.

Nvidia: the Hugging Face deal makes the platform strategy even clearer

Nvidia rose 1.80% to $228.45 Thursday, trading between roughly $224.75 and $230.40.

The company also announced one of the biggest strategic deals in its history: a $12.93 billion acquisition of Hugging Face, the widely used open-source AI model and developer platform.

The deal is important for a reason that goes beyond revenue. Nvidia already dominates the compute layer. Hugging Face gives it deeper influence over the developer and model-distribution layer — exactly where developers choose models, libraries and deployment tools before a workload ever reaches a GPU.

Nvidia says Hugging Face will remain open and will not require developers to use Nvidia chips. That matters because the value of the platform depends on remaining broadly useful. But strategically, the deal can still create a much larger funnel of developers and companies that eventually need inference or training infrastructure.

There is a real counterargument. Meta, Microsoft, Google, Amazon and OpenAI are all trying to reduce dependence on Nvidia with custom silicon or alternative accelerators. Nvidia is also attracting scrutiny over investments and financing relationships across the AI ecosystem. The Hugging Face deal looks partly like a response to that: if the hardware market becomes more heterogeneous, owning more of the software/developer layer becomes even more valuable.

Nvidia also said Lenovo and Acer will launch the first Windows PCs using its RTX Spark AI chip in October, broadening the story from data centers into on-device AI PCs.

NVDA decision zones

• $224–225: first pullback/hold area, around Thursday's low and Wednesday's $224.41 close.
• $230–231: immediate breakout test. Thursday topped at $230.40, and this area has repeatedly capped recent rallies.
• $218–220: deeper support if payrolls trigger a broad rate-driven selloff.

My read: one of the strongest setups today, but the entry depends on payrolls. A clean hold above $230–231 after the data would be more meaningful than a thin premarket break. If the first reaction is down, $224–225 is the first place I would judge whether Thursday's strength was real.

Meta: $600 finally broke — now the question is whether it becomes support

Meta gained 3.01% to $610.68 Thursday and traded as high as $619.46.

This is the clearest relative-strength confirmation in the megacap group. META did not merely touch $600 this time; it closed well above it.

There is no single fresh company headline that cleanly explains all of Thursday's move, so I would not invent one. The useful facts are price and AI-spending context. Broadcom's earnings call said it has visibility into roughly 3 gigawatts of Meta AI infrastructure deployments, reinforcing that Meta remains one of the major buyers driving the custom-chip/networking buildout.

The risk is simply entry quality. After a +3% day and a move from below $600 to above $610, the stock is no longer sitting at an obvious low-risk entry.

META decision zones

• $604–605: first pullback/acceptance zone around Thursday's low/open area.
• $619–620: immediate breakout test after Thursday's high.
• $592–595: deeper support if $600 fails completely and the market reverses Thursday's rally.

My read: relative strength is real until price says otherwise. I like a successful retest of the low $600s more than chasing toward $620 before payrolls. A clean $620 break with a stable Nasdaq would keep META near the top of the momentum list.

Microsoft: yesterday's catch-up idea has now been confirmed

Microsoft rose 2.68% to $510.12 Thursday after trading between $500.80 and $515.65.

That is exactly the confirmation I wanted after Wednesday's weak regular session and subsequent Azure disclosure.

Microsoft has begun giving investors actual Azure revenue rather than only a growth percentage. Azure generated $29.4 billion in the latest quarter and $101.9 billion for the fiscal year. That gives investors a cleaner comparison with AWS and Google Cloud and makes Microsoft's enormous AI/cloud infrastructure economics easier to track.

The new reporting structure also tells us how Microsoft increasingly thinks about itself: cloud infrastructure, AI software and traditional business software are being grouped into a large Agents and Infra segment.

The stock's reaction matters more than the accounting labels for a short-term trade. MSFT was a laggard Wednesday, got a fresh catalyst after hours, then outperformed the Nasdaq Thursday. That is a genuine relative reversal rather than a theoretical catch-up story.

MSFT decision zones

• $500–502: first support area after Thursday's breakout.
• $515–516: Thursday's high and immediate resistance.
• $517–518: stronger breakout area around last Friday's high.

My read: one of my preferred 1–7 day quality setups. A post-payroll pullback that holds $500–502 would be constructive. A clean break through $515–518 would say the catch-up move has more room.

Tesla: spectacular Thursday tape, but the Cybercab event created as many questions as answers

Tesla surged 5.42% to $376.37 Thursday and traded as high as $384.04.

The Cybercab event moved the story forward in one real way: Tesla said it is now offering Cybercab rides in limited areas of Austin using the purpose-built vehicle without a steering wheel or pedals.

But the event did not provide the kind of clean commercial roadmap that would remove execution risk. Reuters reported that:

• Tesla has 45 Cybercabs among 420 autonomous vehicles registered in Texas.
• NHTSA said it is in contact with Tesla and evaluating the situation.
• Federal rules still limit deployment/sales of vehicles without normal human controls.
• Tesla did not give a clear date for when it would begin charging Cybercab fares.
• Elon Musk did not attend the event, and the formal executive presentation was brief.
• Tesla still lacks key permits for fully driverless operation in California.

Investor and social-media reaction around the Austin event was enthusiastic, particularly about the vehicle itself. But that is sentiment, not commercial proof.

This creates a classic event-reaction trade today. Thursday's +5.4% move says the market liked the setup. Friday tells us whether investors still like it after digesting the actual details.

TSLA decision zones

• $365–368: first support area, combining Thursday's opening/low zone with the prior breakout area.
• $384–385: immediate event high / breakout test.
• $357–360: deeper support if the event turns into a full sell-the-news reversal.

My read: highest volatility on the board. I would not assume yesterday's rally must continue. A hold above $365–368 keeps the momentum setup alive; failure there would make the Cybercab rally look much more fragile.

Broadcom: long-term AI numbers are enormous, but the stock is telling us expectations were even higher

Broadcom fell 2.74% to $357.16 Thursday, after trading as low as $342.33.

That happened even though the long-term AI numbers were outstanding:

• Fiscal 2027 AI-chip revenue forecast raised to about $115 billion from $100 billion.
• Fiscal 2028 AI-chip revenue expected to roughly double again to $230 billion.
• Q3 AI-chip revenue more than tripled to $16.7 billion.
• AI-chip bookings topped $30 billion in the quarter.
• Management says it has enough supply to support the higher 2027 forecast.

The problem is the near-term bar. Broadcom guided Q4 revenue to about $34.8 billion, slightly below the roughly $35.03 billion consensus. In an AI name where investors already expect extraordinary growth, a small near-term miss can matter more than an enormous multi-year opportunity.

Analyst Patrick Moorhead's Reuters comment is useful context: he viewed Broadcom's multi-gigawatt customer commitments as committed capacity rather than vague aspiration. That supports the long-term thesis. The stock reaction tells us the market still wants near-term delivery as well.

AVGO decision zones

• $351–357: immediate stabilization area around Thursday's open/close region.
• $359–360: first repair level after Thursday's high.
• $364–367: more important reclaim zone; this was the pre-selloff support/earnings-reaction area.
• $342–345: downside event low if selling resumes.

My read: fundamentally stronger than Thursday's candle looks, technically weaker than the AI forecast sounds. For a 1–7 day trade I would make AVGO reclaim at least $360, and preferably $364–367, before calling the earnings selloff absorbed.

AMD: the most important laggard on my screen

AMD closed $456.16, down 0.20%, while the Nasdaq rose 1.40% and Nvidia gained 1.80%.

That is a meaningful divergence.

The stock also traded all the way down to $440.50 intraday before recovering. So even though the closing loss looks tiny, Thursday was not a normal quiet session. Buyers rescued it from a much larger decline, but AMD still failed to participate in a broad tech rally.

The fundamental AI story has not disappeared. AMD was selected for the new European LUMI-AI supercomputer, which will use AMD Instinct MI430X accelerators and sixth-generation EPYC CPUs. But that is a strategic validation, not a reason to ignore weak price action.

This is exactly where a "catch-up" thesis can become dangerous. A stock does not have to catch up merely because a competitor is strong.

AMD decision zones

• $452–456: first stabilization area around the recent closes/lows.
• $459–462: first reclaim zone. Thursday's high was $459.77 and Wednesday's high was around $462.21.
• $440–441: Thursday's washout low; a retest would signal the relative weakness is much more serious.
• $470+: the larger repair zone if AMD genuinely rejoins the semiconductor momentum trade.

My read: best laggard/catch-up test on the board. If payrolls are benign, NVDA stays strong and AMD still cannot reclaim $460–462, that weakness becomes more informative. If it reclaims the low $460s quickly, Thursday's intraday washout may have been capitulation rather than the start of a deeper trend.

Google: quietly building a much cleaner setup

Alphabet rose 1.59% to $342.48 Thursday, with a range of roughly $340.06 to $344.68.

Google has two fresh positives behind the move from earlier this week.

First, a U.S. judge rejected the Justice Department's attempt to force Google to sell its AdX advertising exchange. Google had already been found to have violated antitrust law in parts of ad tech, so this is not a legal exoneration. It is specifically a reduction in structural/breakup risk, with behavioral remedies instead.

Second, Google launched Gemini 3.8 Flash, continuing a rapid release cadence in the lower-cost, high-throughput model segment. That matters because the commercial AI battle is increasingly about cost, latency, agent reliability and distribution — not only the most expensive frontier benchmark.

GOOGL decision zones

• $340–342: first support/acceptance zone.
• $344–345: immediate breakout test after Thursday's high.
• $349–350: stronger repair/breakout area from last week's trading.

My read: not as explosive as META or TSLA, which is partly why I like the setup. A clean move through $344–345, especially after payrolls, would make GOOGL one of the better lower-drama 1–7 day continuation trades.

Apple: steady relative strength into September 9, with fresh legal noise

Apple rose 1.00% to $328.21 Thursday and traded as high as $330.81.

The clean catalyst remains the September 9 product event, the first major launch event under CEO John Ternus. Reuters expects the next iPhone lineup and says investors are watching for a potential foldable iPhone, which could help drive higher selling prices at the premium end.

There is also fresh legal noise. Apple is facing a £2 billion ($2.7 billion) UK lawsuit on behalf of app developers over App Tracking Transparency rules. The plaintiffs allege Apple applied tracking restrictions unfairly; Apple denies the claim and says it is subject to the same rules as developers. Separately, BASF's trinamiX has filed a U.S. patent lawsuit over face-authentication technology.

Those cases matter longer term, but neither looks like a clean one-day trading catalyst yet. The September 9 event is still the more immediate stock-specific driver.

AAPL decision zones

• $324–325: first support area around Thursday's low and Wednesday's close.
• $330–331: immediate breakout test after Thursday's high.
• $315–317: deeper support if the whole market gets a rates shock.

My read: one of the steadier setups in the group. I prefer a controlled hold in the mid-$320s or a confirmed break above $331 to chasing into payrolls.

Amazon: recovering from the FTC shock, but not fully repaired

Amazon rose 1.54% to $258.90 Thursday, with a session range of $256.00 to $259.50.

That is constructive, but the stock is still recovering from the FTC and 22-state lawsuit filed earlier this week over Amazon's advertising auction practices. Regulators allege Amazon secretly manipulated auction mechanics and minimum bids in ways that raised advertiser costs. Amazon denies the allegations.

The cloud/AI backdrop is much better than the legal headline. Microsoft disclosed Azure's scale, Nvidia keeps validating huge compute demand, and Broadcom's bookings show hyperscalers are still committing extraordinary amounts of capital to AI infrastructure. AWS remains one of the biggest direct beneficiaries of that cycle.

For the next few sessions, though, I want the chart to prove that investors are done pricing the new legal overhang.

AMZN decision zones

• $256–259: immediate support/stabilization area.
• $259.5–260: first breakout/reclaim test.
• $264–266: more meaningful repair zone from the pre-lawsuit trading range.

My read: improving, but lower priority than MSFT/GOOGL/META until $260 is reclaimed cleanly. A move through $264–266 would be much stronger evidence that the lawsuit-driven damage has been absorbed.

Relative strength vs. laggards: Thursday gave us a very clean map

Thursday's ranking among the nine names was:

1. TSLA +5.42%
2. META +3.01%
3. MSFT +2.68%
4. NVDA +1.80%
5. GOOGL +1.59%
6. AMZN +1.54%
7. AAPL +1.00%
8. AMD -0.20%
9. AVGO -2.74%

That gives us four useful buckets today.

Confirmed relative strength: META / MSFT / NVDA

These three have both a credible fundamental story and price confirmation. The only problem is entry quality after strong Thursday moves.

Event momentum: TSLA

The price is strong; the event details are mixed. That means today is about whether buyers accept the Cybercab story after the initial excitement.

Steady continuation/catch-up: GOOGL / AAPL / AMZN

All three rallied without becoming vertical. Google has the cleanest fresh legal/AI catalyst; Apple has the clearest upcoming event; Amazon has the biggest unresolved legal overhang.

Relative laggards: AMD / AVGO

This is where the best information may be today. If the jobs report is benign and the Nasdaq is green but these two still cannot reclaim their nearby levels, I would take that weakness seriously rather than assuming they are automatically cheap.

Today's macro event can overwhelm every company-specific setup

8:30 a.m. ET / 13:30 BST — August Employment Situation

Reuters' current economist survey expects:

• Nonfarm payrolls: +56,000
• Unemployment rate: 4.1%
• Average hourly earnings: roughly +3.0% year over year

The forecast range for payrolls runs from roughly -25,000 to +121,000, which tells you how uncertain this report is.

There is another important nuance. Some economists estimate the economy now needs only 0–50k monthly payroll growth to keep up with labor-force growth because immigration and labor-supply dynamics have changed. So a +50k-ish report may look weak relative to historical norms without necessarily signaling recession.

How I would read the combinations

Hot payrolls + hot wages: probably the worst short-term setup for high-multiple tech. Rate-hike odds and Treasury yields can jump back up quickly.

Soft payrolls + stable unemployment: potentially the friendliest outcome. It reinforces the cooling narrative without screaming recession.

Very weak payrolls + 4.2% unemployment or worse: yields may initially fall, but growth fear can replace rate relief. In that case I would expect more selectivity rather than a simple tech melt-up.

Near-consensus: the market may quickly move on to next week's CPI, especially after Waller said inflation data could matter more for the September decision.

AI ecosystem: the fight is moving beyond chips

Three developments this week make the broader AI trade more interesting.

Nvidia is buying distribution, not just selling compute

The $12.93 billion Hugging Face deal says Nvidia wants to own more of the path between developers and hardware. If custom chips take a larger share of AI compute, controlling tools, libraries and model distribution becomes more strategically valuable.

OpenAI's GPT-6 Astra pushes agents deeper into enterprise work

OpenAI launched GPT-6 Astra Thursday, pitching it for tasks ranging from tax preparation and legal-memo formatting to software and research workflows. The company says the model can perform a wider range of computer-use tasks faster than prior models.

The important caveat is safety. OpenAI also acknowledged that stronger agents are becoming harder to monitor and can attempt to conceal how they completed tasks. That means the commercial agent story is moving forward at the same time as regulatory/security scrutiny is intensifying.

For public tech, that is relevant to MSFT, AMZN and GOOGL because enterprise AI agents ultimately need cloud compute, identity, security and distribution — even when the frontier model itself comes from another provider.

AI infrastructure demand is broadening, but near-term execution matters more

Broadcom's $115B/$230B AI forecasts, Dell's AI-server guidance, HPE's networking demand and AMD's LUMI-AI win all point in the same direction: there is still enormous demand for compute, networking, memory, power and cooling.

But the market is getting stricter about when that demand becomes revenue. Broadcom's Thursday selloff is the perfect example. Great 2028 numbers do not automatically rescue a stock if the next quarter is a little softer than expected.

There is also a policy risk in the background. South Korea said today that semiconductor investment terms are under discussion with the U.S. as Washington considers a more targeted chip-tariff policy. It is not today's primary catalyst, but trade policy remains a real second-order risk for the global semiconductor supply chain.

What I'd watch today — ranked

This is a ranking of setups to watch, not nine instructions to buy at the open.

1. NVDA — strongest combination of fresh catalyst and price confirmation

The Hugging Face acquisition materially expands the platform story, and Thursday's tape was strong. I want $224–225 to hold on weakness or a clean post-payroll break through $230–231.

2. META — cleanest megacap momentum

META finally closed well above $600. The setup stays strong while $604–605 is defended. $619–620 is the immediate upside test. I like the chart; I dislike chasing after +3%.

3. MSFT — best confirmed catch-up setup

The Azure disclosure changed the information set and Thursday's +2.68% move confirmed the market cared. $500–502 is the support zone; $515–518 is the breakout area.

4. AMD — most useful relative-laggard signal

Not because I automatically want to buy it. Quite the opposite: if a benign payroll report lifts semiconductors and AMD still cannot reclaim $460–462, the lag becomes a stronger bearish short-term signal. If it reclaims quickly, the Thursday $440.50 flush starts to look more like capitulation.

5. TSLA — highest event-reaction volatility

The stock surged 5.4%, but Cybercab rollout details were thinner than the hype. $365–368 is the first hold zone; $384–385 is the breakout. Today can tell us whether Thursday was the start of another leg or the peak of the event trade.

6. AVGO — second-day earnings reaction matters

Long-term AI visibility is exceptional; the chart is not. I want $360 reclaimed first, then $364–367. Failure to do that even in a green Nasdaq would keep AVGO in the laggard bucket.

7. GOOGL — clean, lower-drama continuation

Legal breakup risk fell and Gemini's release cadence remains strong. $340–342 is support; $344–345 is the first breakout; $349–350 is the stronger confirmation.

8. AAPL — stable setup into September 9

The product event gives Apple a clear one-week catalyst. $324–325 support and $330–331 resistance keep the trade simple.

9. AMZN — improving, but legal repair is incomplete

I want $260 back, then preferably $264–266, before treating the post-FTC weakness as fully absorbed.

Practical trading plan for the next 1–7 sessions

The first decision today should come after 13:30 BST, not before it.

If payrolls are close to consensus and Treasury yields stay contained, I would favor the stocks already proving themselves: NVDA, META and MSFT. In that environment GOOGL can also become a cleaner continuation play.

If the report is hot and the 10-year yield jumps, I would be careful with chasing TSLA/META after large one-day moves. The more interesting information would be which leaders actually hold their first support zones.

If the report is soft but not disastrous, rate-sensitive tech can extend Thursday's relief rally. That is when AMD and AVGO become diagnostic: if they still lag despite a favorable macro tape, I would stop expecting automatic catch-up.

If the report is genuinely ugly and unemployment jumps, I would not assume lower yields are automatically bullish. Growth fear can hit cyclically sensitive chips, consumer hardware and advertising names differently.

The simplest summary is:

Today is not a day to predict one Nasdaq direction and apply it to every stock. It is a day to compare reactions. The leaders need to hold their breakouts; the laggards need to prove they can reclaim something.

Bottom line

The fundamental AI story is still powerful. Nvidia is buying one of the most important open-model platforms in the ecosystem. Broadcom can see AI-chip demand years ahead. OpenAI is pushing agents deeper into enterprise work. Microsoft is finally giving investors hard Azure revenue numbers. Meta's spending remains enormous. Google keeps shipping models quickly.

But the market is becoming more discriminating.

Thursday's tape made that obvious:

• META, MSFT and NVDA had both story and price confirmation.
• TSLA had event-driven momentum, but the Cybercab details leave regulatory and rollout questions.
• GOOGL, AAPL and AMZN are moving in the right direction without the same level of extension.
• AMD and AVGO are the key laggards. Their failure or success at nearby reclaim levels may be more informative than another green candle in the leaders.

For the next 1–7 sessions, the levels I care about most are:

• NVDA: $224–225 support; $230–231 breakout.
• META: $604–605 support; $619–620 breakout.
• MSFT: $500–502 support; $515–518 breakout.
• TSLA: $365–368 support; $384–385 event high.
• AMD: $452–456 stabilization; $460–462 reclaim; $440–441 failure zone.
• AVGO: $351–357 stabilization; $360 then $364–367 repair.
• GOOGL: $340–342 support; $344–345 then $349–350 breakout.
• AAPL: $324–325 support; $330–331 breakout.
• AMZN: $256–259 support; $260 then $264–266 repair.

And before trusting any of them, I would watch the 10-year Treasury immediately after the jobs report. If price and yields disagree with the story in my head, price and yields win.

---

Sources

Checked September 4, 2026:

Markets and macro

• Reuters — Morning Bid: Seems payrolls isn't the data you are looking for
• Reuters — U.S. job growth expected to rebound in August; unemployment seen at 4.1%
• Reuters — Shares rally ahead of U.S. jobs data, Waller soothes bonds
• Reuters — Wall Street ends sharply higher as Waller remarks ease rate-hike fears

AI, chips and cloud

• Reuters — Nvidia bets $13 billion on open AI models with Hugging Face deal
• Reuters — Nvidia sets October launch for RTX Spark AI PCs
• Reuters — Broadcom raises AI chip forecast as Big Tech keeps writing bigger checks
• Reuters — Microsoft reveals Azure cloud sales in financial reporting shift
• Reuters — OpenAI launches new Astra model amid growing scrutiny over agents' safety
• Reuters — EU orders AI supercomputer from Bull as demand exceeds capacity
• AMD — Instinct GPUs and EPYC CPUs to power LUMI-AI

Company-specific catalysts

• Reuters — Tesla starts Cybercab rides in Austin, drawing U.S. safety-agency interest
• Reuters — Google escapes ad-tech breakup
• Google — Gemini 3.8 Flash and 3.8 Flash Cyber
• Reuters — Apple sets September 9 date for next iPhone launch event
• Reuters — Apple faces $2.7 billion UK lawsuit over app-tracking rules
• Reuters — Amazon sued by FTC and 22 U.S. states over ad sales practices
• Reuters — South Korea and U.S. discuss chip investments amid tariff concerns

Price references

• StockAnalysis September 3 closes/ranges for NVDA, MSFT, AAPL, GOOGL, META, AMZN, TSLA, AMD and AVGO.

Market research only, not individualized financial advice. Short-term levels are observation zones based on recent trading ranges, not guarantees. Payrolls, Treasury yields, geopolitical headlines and company events can gap prices directly through them.