Daily Tech Report

Research cutoff: Thursday, September 3, 2026, roughly 08:30 BST.
U.S. price cutoff: Wednesday, September 2 regular-session close, plus confirmed Wednesday after-hours reactions where explicitly noted.
Horizon: roughly the next 1–7 U.S. trading sessions.
Pre-market note: It is still only about 03:30 ET, before the most liquid individual-stock U.S. premarket. Reuters' latest global-market snapshot has S&P 500 e-mini futures up about 0.07%, so I am using Wednesday's confirmed closes and after-hours moves rather than pretending thin overnight stock quotes are a reliable opening signal.

The quick read

Thursday starts with a slightly better market tone than the last few mornings, but I would not call the macro problem solved.

U.S. and Japanese bond yields are edging lower after the recent global selloff, and S&P 500 futures are fractionally positive. The problem is that oil remains above $90 a barrel as U.S.-Iran strikes continue, while markets now price roughly a 67% probability of a 25-basis-point Fed hike in September, up from 37% a week ago according to Reuters' summary of CME FedWatch.

Against that backdrop, the corporate AI news is unusually strong:

• Broadcom raised its fiscal 2027 AI-chip revenue forecast to roughly $115 billion from $100 billion and expects about $230 billion in fiscal 2028. That is one of the clearest signs yet that AI infrastructure spending is broadening beyond Nvidia GPUs into custom accelerators and networking.
• The catch: Broadcom's Q4 revenue guide of about $34.8 billion came in slightly below the roughly $35.03 billion Wall Street estimate, and the stock traded lower after hours. Great long-term AI visibility did not automatically produce a great immediate stock reaction.
• Nvidia rose 3.2% Wednesday, leading a semiconductor rebound after the group had been hit hard over the previous several weeks.
• Microsoft fell 0.84% in the regular session, then rose about 1.4% after hours after disclosing Azure revenue for the first time: $29.4 billion in the latest quarter and $101.9 billion for the fiscal year.
• Google got two company-specific positives: it avoided a forced breakup of its AdX advertising exchange, and it launched Gemini 3.8 Flash, which Google describes as its strongest Flash-tier model for coding, agents and complex workflows.
• Tesla's Cybercab event is today in Austin. That makes TSLA the highest-event-risk name on the board, but Tesla still has not disclosed exactly what kind of launch it is, the timing, or whether commercial regulatory approvals are in place.

Wednesday's tape also gave us a useful split. NVDA and META showed real relative strength. MSFT and AMD lagged. AMZN was essentially flat. That means I would not use one blanket "tech bullish" view today.

My working view is:

The AI-capex story looks stronger after Broadcom, but the macro discount rate is still hostile. I want to own or trade the names where price is confirming the story — and treat the laggards as catch-up candidates only after they actually reclaim something.

Market pulse before the U.S. open

Wednesday itself was broader than the index move suggests. Reuters reported the Russell 2000 up 1.1%, while Nasdaq advancers outnumbered decliners by almost two to one. But software and services underperformed while semiconductors rose.

That is important for today's relative-strength screen: the market was willing to re-risk, but it preferred chips and selected large caps rather than buying every technology subsector equally.

The tech board

The price zones below are observation/decision areas derived from recent daily ranges. They are not analyst targets, guarantees or automatic buy/sell levels.

Broadcom: the AI demand signal was excellent — the stock reaction was not

Broadcom is the most important read-through for today's AI trade.

The headline numbers were strong:

• 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 sales more than tripled to $16.7 billion.
• Total Q3 revenue was $29.59 billion, ahead of the roughly $29.36 billion consensus.
• Adjusted EPS was $3.32, above the roughly $3.24 estimate.
• AI-chip bookings topped $30 billion in the quarter.

CEO Hock Tan also gave unusually concrete infrastructure visibility through 2028, including more than 10 gigawatts tied to Anthropic, more than 5 GW for OpenAI and 3 GW for Meta. Analyst Patrick Moorhead's read was essentially that this looks like committed capacity rather than vague aspiration — a useful distinction because investors have become much more skeptical of AI pipelines whose revenue is years away.

So why did AVGO fall after hours?

The near-term guide was a little light. Broadcom expects about $34.8 billion of Q4 revenue, versus an LSEG consensus around $35.03 billion. The shares fell more than 3% at one point in extended trading before recovering much of the drop; StockAnalysis' last confirmed after-hours quote was around $364.23, down 0.82% from the regular close.

That reaction is a good reminder: the market already expects a lot from AI winners. A huge long-term opportunity does not guarantee a positive one-day earnings reaction if the immediate quarter is merely good instead of spectacular.

AVGO decision zones

• $364–367: immediate post-earnings reaction zone, combining Wednesday's low/close and the confirmed after-hours area.
• $371–373: first reclaim zone. Getting back above Wednesday's high area would show the market is digesting the soft Q4 guide constructively.
• $376–377: bigger breakout area from last week's range.

My read: fundamentally, the report strengthened the AI-infrastructure thesis. For a 1–7 day trade, I would let the opening reaction settle. If AVGO cannot reclaim the low $370s despite the raised 2027/2028 AI forecasts, that is useful information about how demanding the valuation bar has become.

Nvidia: Wednesday's rebound now has a Broadcom confirmation behind it

Nvidia rose 3.21% to $224.41 Wednesday, with a session range of roughly $218.48 to $227.95.

That was one of the strongest moves in the megacap tech group and came while the semiconductor index itself was rebounding from a rough multi-week stretch.

Broadcom's earnings are not a direct Nvidia revenue forecast, but they are a powerful cross-check on the same underlying demand cycle. Hyperscalers and AI labs are still committing enormous amounts of capital to compute, networking and custom accelerators. The fact that custom silicon is expanding does not automatically mean Nvidia loses; for now, it looks more like the total AI infrastructure pie is still expanding rapidly.

The risk remains macro. NVDA is a high-beta name, and if today's ISM Services data or Waller's comments push Treasury yields sharply higher again, a great AI narrative can still get sold.

NVDA decision zones

• $218–220: first pullback/hold area, around Wednesday's open/low and the prior close.
• $227–228: immediate breakout test after Wednesday's high.
• $216–217: deeper support if the first zone fails; that area was repeatedly defended in the previous few sessions.

My read: one of today's better relative-strength setups. I prefer either a controlled pullback that holds $218–220 or a clean acceptance above $228. After a 3.2% day, I do not need to chase the first green tick.

Microsoft: weak regular session, but the Azure disclosure changes today's setup

Microsoft closed $496.82, down 0.84% Wednesday, making it one of the weaker large-cap names during the regular session.

Then came a useful after-hours catalyst.

Microsoft disclosed Azure revenue on a quarterly basis for the first time. Azure generated $29.4 billion in the latest quarter and $101.9 billion for the fiscal year ended June 30. That gives investors a much cleaner comparison across the cloud leaders: Reuters noted AWS reported $42.2 billion in its latest quarter, while Google Cloud reported $24.8 billion.

Microsoft is also reorganizing reporting into two major segments, including an Agents and Infra segment that will group cloud computing, AI software and traditional business software. The company slightly lowered the stated Azure growth forecast because some GitHub revenue is being reclassified into Microsoft 365 Cloud, but said the overall quarterly outlook is unchanged.

MSFT shares rose about 1.4% after hours after the disclosure.

That makes Microsoft a classic catch-up watch today: Wednesday's regular-session tape was weak, but there is now fresh information that could change the next session's relative behavior.

MSFT decision zones

• $494–497: immediate support/decision area; Wednesday's low was $493.81 and the close was $496.82.
• $500: first repair level. It is psychologically simple and roughly Wednesday's intraday high.
• $505–506: more important reclaim area from Tuesday's range.

My read: if MSFT opens firmer and holds above $500 while the Nasdaq is stable, the after-hours catalyst may be turning a laggard into a short-term catch-up trade. If it loses $494 quickly, I would not force the thesis.

Google: one legal win plus one fresh AI-product catalyst

Alphabet rose 0.63% to $337.12 Wednesday.

The legal headline was meaningful even though it does not erase Google's broader antitrust risk. A U.S. judge rejected the Justice Department's request to force Google to sell its AdX advertising exchange. The judge instead accepted behavioral remedies. Google had previously been found to have illegally maintained monopolies in parts of ad tech, so this is not a clean legal exoneration — it is specifically a win on the remedy and breakup risk.

The same day, Google launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber. Google says 3.8 Flash improves on 3.7 across software engineering, agentic tasks and multi-step reasoning while keeping the same introductory API price: $0.75 per million input tokens and $3.75 per million output tokens.

For investors, the interesting part is not a benchmark leaderboard by itself. It is Google's release cadence. This is its third Flash release in six weeks, which tells us Google is pushing hard on the lower-cost, high-throughput part of the model market where real enterprise workloads can scale.

GOOGL decision zones

• $333–335: first support area, around Wednesday's low and Tuesday's close.
• $340: immediate breakout test; Wednesday topped out exactly around that level.
• $344–347: next repair zone from the previous week's trading.

My read: improving. The breakup risk just fell, the AI product cadence is strong, and the stock actually rose Wednesday. I would become more constructive on a clean $340 break rather than buying purely because the legal headline sounds positive.

Meta: Wednesday's cleanest relative-strength signal

Meta rose 2.47% to $592.85, with an intraday high of $600.38.

That is a strong move relative to both the Nasdaq and most of the megacap group.

There was no single new Meta headline that cleanly explains the whole 2.5% move, and I would not invent one. The more useful interpretation is that investors were willing to buy Meta aggressively as risk appetite improved, while Broadcom's report after the close reinforced the scale of AI infrastructure commitments from customers including Meta.

The stock is now approaching a very obvious psychological level.

META decision zones

• $577–580: first pullback support, roughly Wednesday's low / Tuesday close area.
• $592–595: near-term acceptance zone around Wednesday's close.
• $600–601: immediate breakout test.

My read: strongest price confirmation in the group, but worse entry quality after a +2.5% day. A clean hold above $600 would be momentum confirmation; a pullback toward the high $570s would offer a much clearer risk point.

Apple: stable tape while the September 9 catalyst gets closer

Apple finished almost unchanged at $324.96, after trading between $323.53 and $328.40.

That is not exciting price action, but Apple has a clear catalyst path.

John Ternus has now taken over as CEO, and Apple's next major product event is scheduled for September 9. Reuters expects the next iPhone lineup to be the focus, with investors also watching for a potential foldable iPhone and for signs that the new leadership can close Apple's perceived AI gap.

The stock's current setup is simpler than the narrative: it is consolidating in the low-to-mid $320s ahead of a known event.

AAPL decision zones

• $323–325: immediate support / consolidation zone.
• $328–330: first breakout area.
• $315–317: deeper support if the market gets a rate-driven selloff.

My read: lower urgency than NVDA, AVGO or TSLA today, but one of the cleaner one-week catalyst setups. I would rather buy a controlled hold than chase right before September 9.

Tesla: today's biggest event risk

Tesla closed $357.01, up 0.26% Wednesday, after trading between roughly $349.92 and $360.62.

The Cybercab event in Austin is today. That alone makes normal support/resistance less reliable than usual because an event headline can gap the stock through several levels immediately.

Reuters confirms a few things and, importantly, leaves several things unresolved:

• Cybercab is Tesla's purpose-built autonomous two-seater without steering wheel or pedals in its intended form.
• Tesla has already produced some units and has been testing them on public roads.
• The company invited current robotaxi riders and selected guests to an Austin launch event.
• Tesla has not announced the event time or clearly defined what kind of launch this is.
• Tesla has not said whether it has all approvals needed for commercial Cybercab service or whether the vehicle meets federal safety requirements for that use.

So today is not simply "Cybercab launch = bullish." The stock can react very differently depending on whether Tesla announces a concrete commercial rollout, a limited employee/guest program, production targets, regulatory milestones or mostly a showcase.

TSLA decision zones

• $350–353: first support area, around Wednesday's low and recent two-day lows.
• $360–361: immediate breakout/reclaim zone.
• $368–369: the more important event breakout area, around Monday's post-rally high.

My read: potentially the most volatile stock today and therefore not the one I would front-run casually. I want the event to tell us whether the market accepts the story above $360–361 or sells it back toward $350.

AMD: still the cleanest laggard/catch-up test

AMD fell 0.55% to $457.06 Wednesday while Nvidia gained more than 3%.

That is meaningful relative weakness.

The fundamental AI backdrop has not become worse. AMD recently won a role in Europe's new LUMI-AI supercomputer project, and Broadcom's results reinforce that AI infrastructure spending remains very strong across the stack. But a catch-up thesis only becomes tradable when the price starts confirming it.

Wednesday's range was roughly $452.36 to $462.21, almost identical on the downside to Tuesday's low around $452.30. That gives us a very clean short-term decision area.

AMD decision zones

• $452–457: immediate stabilization zone.
• $462: first reclaim level.
• $470–476: bigger repair zone from Monday/Friday trading.

My read: this is the name where I most want to avoid saying "it has to catch up." If AMD holds the low $450s and reclaims $462 while NVDA/AVGO stay firm, the relative-value setup improves quickly. If $452 breaks, the market is telling us the lag is real.

Amazon: stabilizing, but the legal overhang has not disappeared

Amazon finished at $254.98, essentially flat, with a tight Wednesday range of $253.40 to $256.24.

That is at least more stable than Monday and Tuesday, when the stock was digesting the FTC and 22-state lawsuit over advertising sales practices. Amazon denies the allegations, and they remain allegations rather than established findings, but the case matters because advertising is a high-margin business.

The broader cloud read-through is better. Snowflake raised its annual product-revenue forecast Wednesday on strong cloud and AI demand, and the company has a large multiyear AWS relationship. Microsoft also gave investors a cleaner look at the sheer size of the cloud market by disclosing Azure sales.

So Amazon has a mixed short-term setup: the AWS/AI environment looks healthy, but the stock itself has not recovered from the fresh regulatory shock.

AMZN decision zones

• $253–255: immediate support/stabilization zone.
• $259–260: first reclaim area.
• $264–266: more meaningful repair zone from last week's breakout.

My read: a possible catch-up, but lower priority than MSFT because Amazon still has an unresolved company-specific legal overhang. I want $259–260 back before getting more aggressive.

Relative strength: what Wednesday actually told us

Among the nine names, Wednesday's regular-session ranking was roughly:

1. NVDA +3.21%
2. META +2.47%
3. GOOGL +0.63%
4. TSLA +0.26%
5. AMZN +0.02%
6. AAPL -0.05%
7. AMD -0.55%
8. AVGO -0.66%
9. MSFT -0.84%

But that ranking needs one important adjustment: MSFT rose about 1.4% after hours after the Azure disclosure, while AVGO traded lower after earnings.

That leaves four useful buckets for today:

• NVDA / META: confirmed relative-strength leaders. Better momentum, but worse chase risk after strong Wednesday moves.
• GOOGL: improving relative strength with two fresh company-specific catalysts.
• MSFT / AMD / AMZN: catch-up candidates. MSFT has the strongest fresh catalyst; AMD has the cleanest technical decision point; Amazon has the biggest legal overhang.
• AVGO / TSLA: event-driven names where earnings/event reactions matter more than yesterday's normal chart.

The biggest relative-value idea for me is MSFT versus its own Wednesday weakness, not blindly buying the weakest stock in the table. The Azure disclosure gives us a real reason the relative trend could change today.

Today's macro calendar can still overpower the company news

The corporate setup is bullish enough that it would be easy to ignore the rates story. I would not.

8:30 a.m. ET / 13:30 BST — productivity/labor costs and Fed Governor Waller

The Bureau of Labor Statistics releases the revised second-quarter Productivity and Costs report at 8:30 a.m. ET.

At almost exactly the same time, Fed Governor Christopher Waller is scheduled to speak at Reuters NEXT. Reuters notes that Waller said in July that higher rates might be needed in the near term.

That combination can move yields before the equity open. Unit labor costs matter because stronger labor-cost pressure can reinforce the inflation argument for another hike.

10:00 a.m. ET / 15:00 BST — ISM Services

ISM releases its August Services PMI at 10:00 a.m. ET.

Services matter more than manufacturing for the U.S. economy, and the inflation/employment components can move the Fed narrative quickly.

The awkward setup for tech is:

• Hot services + hot prices: yields can rise and high-duration tech can sell even with strong AI fundamentals.
• Very weak services: yields may fall, but investors can start worrying about growth.
• Moderate services: probably the cleanest outcome for megacap tech.

Friday, September 4 — the real macro event

Friday's official U.S. employment report remains the biggest macro catalyst of the week.

ADP said private payrolls rose only 38,000 in August, below the Reuters economist consensus of 48,000. But Reuters explicitly notes that ADP has been a poor predictor of the BLS payroll number.

The Reuters survey currently expects:

• Nonfarm payrolls: +56,000
• Unemployment rate: 4.1%, unchanged

With September hike odds already around 67%, Friday can reset the whole rates/tech equation.

AI ecosystem: Broadcom is not the only demand signal

The strongest theme this morning is that AI spending is showing up across multiple layers of the stack rather than only in GPU sales.

Dell jumped strongly Wednesday after raising its annual sales and profit forecasts on demand for AI-optimized servers. That is a real-world server-shipment signal.

Snowflake raised its annual product-revenue forecast after stronger cloud and AI demand. It said new AI products are contributing materially to growth, including thousands of customer accounts using newer AI developer and work products.

HPE also raised forecasts on AI-server and networking demand, though its shares fell after hours as management warned that memory, NAND, CPU and drive supply remains constrained. That is another useful distinction: demand is strong enough that supply can become the problem.

And Broadcom itself now sees roughly $115 billion of AI-chip sales in fiscal 2027 and about $230 billion in fiscal 2028.

Put those together and my interpretation is:

The next question for the AI trade is less "is there demand?" and more "which companies can turn enormous demand into near-term revenue, with enough supply and margins to justify the stock price?"

That is why NVDA can rally while AMD lags, and why Broadcom can give a spectacular long-term forecast yet still fall after hours on a slightly soft near-term guide.

What I'd watch today

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

1. AVGO — most important information event

The long-term AI forecast was stronger than I expected, but the initial stock reaction was negative. I want to see whether $364–367 holds and whether AVGO can reclaim $371–373. If it cannot, the market is telling us that the near-term guide matters more than the huge 2027/2028 opportunity for now.

2. MSFT — best catch-up setup

Wednesday regular-session weakness plus a 1.4% after-hours Azure catalyst is exactly the kind of relative reversal I like to watch. $494–497 is support; $500 is the first repair; $505–506 would make the setup much cleaner.

3. NVDA — best confirmed chip momentum

Wednesday's +3.2% move plus Broadcom's AI-demand confirmation keeps Nvidia near the top of the list. I prefer $218–220 on a controlled pullback or a clean push through $227–228.

4. GOOGL — two fresh catalysts, price beginning to cooperate

Google avoided a forced AdX breakup and released Gemini 3.8 Flash on the same day. $333–335 is the support area and $340 is the first real breakout test.

5. TSLA — highest event volatility

Cybercab is today. I care less about predicting the headline and more about the reaction: $360–361 is the first upside test; $350–353 is the first downside support. A move through $368–369 would be a much stronger momentum signal.

6. META — strongest Wednesday tape, but less attractive entry

META nearly touched $600 after a +2.5% day. A breakout above $600–601 is meaningful, but I would rather not chase it if the macro tape turns shaky.

7. AMD — pure catch-up test

The AI story is fine; the price is not confirming it yet. I want $452–457 to hold and $462 reclaimed before treating it as a serious relative-value trade.

8. AAPL — stable one-week catalyst setup

The September 9 event is close. $323–325 support and $328–330 resistance keep the setup simple.

9. AMZN — wait for legal-news damage to repair

The cloud read-through is good, but AMZN remains below the levels it held before the FTC lawsuit. $253–255 needs to remain a floor, and $259–260 is the first level I want back.

Today's catalyst clock

Before the U.S. open

• 8:30 a.m. ET / 13:30 BST: revised Q2 U.S. Productivity and Costs
• Around 8:30 a.m. ET / 13:30 BST: Fed Governor Christopher Waller at Reuters NEXT

During the U.S. session

• 10:00 a.m. ET / 15:00 BST: August ISM Services PMI
• Tesla Cybercab event: Thursday in Austin; Tesla had not announced an official event time as of the research cutoff

Next major event

• Friday, September 4, 8:30 a.m. ET: August U.S. Employment Situation

The timing matters. Waller and labor-cost data can move yields before the cash open, ISM can change the direction 30 minutes into trading, and Tesla can generate company-specific volatility later regardless of what the Nasdaq is doing.

Bottom line

Thursday has a better corporate backdrop than macro backdrop.

Broadcom just gave us one of the strongest long-range AI infrastructure demand signals of the year. Nvidia rebounded 3.2% Wednesday. Microsoft finally gave investors hard Azure revenue numbers. Google reduced a major breakup risk while launching another fast-moving Gemini model. Dell, Snowflake and HPE are all showing that AI spending is translating into servers, cloud workloads and networking demand.

But the market still has to price all of that with oil above $90, a roughly 67% probability of a September Fed hike, and Friday's jobs report one day away.

So for the next 1–7 sessions, I would separate the board like this:

• AVGO: fantastic AI demand, but make the post-earnings stock prove it can reclaim $371–373.
• MSFT: my favorite catch-up candidate if the Azure disclosure carries the stock back above $500 and then $505–506.
• NVDA: strongest chip momentum; $218–220 pullback support, $227–228 breakout.
• GOOGL: legal + AI catalysts make $340 the key short-term breakout.
• TSLA: event trade; let Cybercab information and the $350/$361 reaction tell us what the market actually believes.
• META: strongest Wednesday tape, but be careful chasing just under $600.
• AMD: $452–457 must hold before the laggard becomes a catch-up trade.
• AAPL: stable setup into September 9; $323–325 support, $328–330 breakout.
• AMZN: still needs to repair the FTC-driven damage; $259–260 is the first reclaim I care about.

The single biggest mistake today would be to see Broadcom's huge AI forecast and assume every AI-related stock must go up immediately.

The better trade is to watch which names actually absorb today's macro data and still hold their levels. That is where the relative strength is real rather than theoretical.

---

Sources

Checked September 3, 2026:

Markets and macro

• Reuters — Morning Bid: Bonds breathe sigh of relief
• Reuters — Wall St ends higher as stocks reclaim some shine
• Reuters — U.S. private payroll growth slows in August
• BLS — September 2026 release calendar
• ISM — August 2026 Services PMI release information

AI, cloud and semiconductors

• Reuters — Broadcom raises AI chip forecast as Big Tech keeps writing bigger checks
• Reuters — Microsoft reveals Azure cloud sales in financial reporting shift
• Google — Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
• Reuters — Dell shares gain after strong AI server demand boosts annual forecast
• Reuters — Snowflake lifts annual revenue forecast on cloud and AI demand
• Reuters — HPE raises forecasts on AI demand, shares fall on supply concerns

Company-specific catalysts

• Reuters — Google escapes ad tech breakup
• Reuters — Tesla set to hold Cybercab event in Austin
• Reuters — Apple sets September 9 date for next iPhone launch event
• Reuters — Amazon sued by FTC and 22 U.S. states over ad sales practices

Price references

• StockAnalysis September 2 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. Earnings, Fed commentary, geopolitical headlines and macro data can gap prices straight through them.