Nvidia reports on Wednesday and the Federal Reserve’s new chair speaks on Friday, and between them they account for most of what this week is about. $NVDA’s July quarter lands after the US close on the 26th against a guide of $91bn give or take 2% – a guide that assumes no data-centre compute revenue from China at all, and that was set before Washington’s H200 clearances began turning into actual shipments. Options are pricing a move of roughly 7%, against an average nearer 2.8% across the past four quarters. Two days later, at 10am Eastern on Friday, Kevin Warsh gives his first Jackson Hole keynote as Fed chair, into a market coming off a losing week – all three US indexes finished lower in the week to 21 August despite Friday’s rebound, with rising Treasury yields, government borrowing and a firmer crude price doing most of the damage. Wednesday evening is triple-booked, because CrowdStrike and Salesforce report alongside Nvidia: the first at roughly 175 times forward earnings after losing about 15% in six sessions, the second having already de-rated about 21% this year. Monday opens with PDD before the US bell, and Thursday carries the largest names in a Canadian bank bloc that runs Tuesday to Thursday. Friday closes on Asia – four Chinese megabanks, PetroChina, Meituan, and BYD’s interim result, the first half in which exports have been large enough to reshape its profit and loss. Twenty-four names make our calendar; Marvell, fresh from granting Google a warrant over as many as 58.97m shares at $206.58 – about $12.2bn if it fully vests and is exercised, rather than stock handed over at signing – plus Intuit, Synopsys, Autodesk, Workday, CNOOC and China Life fill in the rest.
🇺🇸 NVIDIA
$NVDA reports fiscal second-quarter results Wednesday after the US close – the print the rest of the tape is waiting on, and the read-through to every hyperscaler capital-expenditure line behind it. The shares closed Friday at $214.72, down 0.98%, having drifted lower in each of the last six sessions from $225.30 on 13 August; the 52-week range is $164.07 to $236.54, and on trailing earnings the stock trades near 33 times. Options are pricing a post-print move of roughly 7%, well above the roughly 2.8% the stock has actually averaged across the past four quarters.
The April quarter was the base, and it was a large one: revenue of $81.6bn, up 85% year on year and 20% sequentially, with Data Center at $75.2bn, up 92% and 21%. Inside that, compute was $60.4bn, up 77%, and networking $14.8bn, up 199% year on year and 35% sequentially – now close to a fifth of the segment. GAAP and non-GAAP gross margins were 74.9% and 75.0%, GAAP diluted earnings were $2.39 and non-GAAP $1.87, and free cash flow was $48.6bn. The company returned about $20.0bn to shareholders in the quarter, added an $80.0bn repurchase authorisation and lifted the quarterly dividend from $0.01 to $0.25. Guidance for this quarter is $91.0bn of revenue give or take 2%, gross margins of 74.9% GAAP and 75.0% non-GAAP give or take 50 basis points, and GAAP operating expenses near $8.5bn – and it explicitly assumes no Data Center compute revenue from China. Consensus sits fractionally above the midpoint, near $91.7–91.9bn, with non-GAAP earnings around $2.08–2.18. For scale, the year-ago July quarter was $46.7bn of revenue, $41.1bn of Data Center and $1.05 of non-GAAP earnings, at a 72.7% non-GAAP gross margin.
China is the live variable rather than the settled one. Washington has cleared H200 sales to roughly ten Chinese firms under conditional licences, with the per-buyer ceiling reported at 75,000 chips under the framework issued in January and later reporting in August putting it as high as 100,000; Lenovo and Foxconn are named as authorised distributors. Actual volumes are a fraction of either figure. ByteDance and Tencent are each reported to have taken delivery of around 10,000 units in recent weeks – about 13% of the lower ceiling – and Beijing has been discouraging purchases in order to limit dependence on American silicon, even as it clears the imports. Blackwell-class parts remain blocked. None of it is in the guide.
Our readThe October-quarter guide is the swing factor, not the July quarter. At $91.0bn the current guide asks for about 11.5% sequential growth on an $81.6bn base, and consensus already sits on it; what is undecided is the next step up, where the published range of expectations is unusually wide – the market is closer to $104bn while at least one large broker has framed $107–108bn as the bar. That spread, rather than the beat, is the event. Watch gross margin against a 75.0% guide with 50 basis points of stated tolerance, since H200 is older silicon and any China mix arrives at a different price point; Data Center networking, up 199% year on year and increasingly the part of the franchise that competitors have to replicate rather than substitute; and whatever management says about China revenue recognition, given the outlook carries none of it, which makes any approved shipment upside to a number the company chose not to bank – though on the current evidence the binding constraint is at least as much Beijing’s stance as Washington’s licences. The read-across is about as broad as it gets – hyperscaler capex, the AI supply chain, and the two other prints landing the same evening.
🇨🇦 Royal Bank of Canada
RBC reports fiscal third-quarter results Thursday before the Toronto open – the largest of six Canadian banks reporting across three days, and the cleanest developed-market read on whether a consumer credit cycle is turning. Bank of Montreal and Scotiabank go on Tuesday, National Bank on Wednesday, and CIBC, TD and RBC together on Thursday. The New York line closed Friday at $205.23, effectively unchanged on the day but about 5% below its 14 August close, after the Canadian bank group sold off mid-week; the 52-week range is $136.10 to $218.57, and the stock trades near 18 times trailing earnings on a dividend yield around 2.3%. The Toronto line closed at C$282.37.
The April quarter set a high base. Net income was $5.5bn, up 25% year on year, with adjusted diluted earnings of $3.90, up 25%, on $17.45bn of revenue and a 17.2% return on equity. Every operating segment grew: Personal Banking $1,870m, up 17%; Commercial Banking $854m, up 43%; Wealth Management $1,185m, up 28%; Capital Markets $1,484m, up 23%. The credit line is the one worth reading twice, and it needs both comparisons. Total provisions for credit losses fell 36% year on year to $912m, but the year-ago quarter carried a large overlay for trade disruption, and most of that decline sits in the performing-loan line – which was still a provision of $18m rather than a release, $550m below a year earlier and $10m below the prior quarter. Provisions on impaired loans were $899m, up 6% year on year but down $169m, or 16%, sequentially. The provisioning ratio was 35 basis points, down 23 year on year and six from the prior quarter, with the impaired-loan ratio at 34 basis points against 40 sequentially and 35 a year ago. Total allowances stood at $7.8bn and CET1 at 13.5%, and the quarterly dividend was raised 7% to $1.76. Consensus for this quarter is around C$4.04. Context matters for the bar: all six banks have beaten consensus for four consecutive quarters, RBC by an average of about 8.5%, and its earnings estimates have been revised up roughly 9.8% over twelve months.
The macro backdrop has improved rather than deteriorated. July’s Labour Force Survey showed employment up 75,000 and the unemployment rate down a tenth to 6.4%, a third consecutive monthly fall and the lowest since July 2024, with Ontario adding 52,000 of the gain and wages up 2.8% year on year. Housing is the softer half of the picture, though less uniformly than a single number suggests. The Toronto board’s July figures put the average selling price at $1,003,956, down 4.5% year on year, and the MLS Home Price Index composite – a separate measure, built to strip out changes in what actually sold – down 4.6%. Volumes point the other way: sales slipped 0.9% while new listings fell 17.8%, which the board itself characterised as a tightening market.
Our readThe composition of provisions is the swing factor, not the headline number. The performing-loan provision of $18m is unusually low, but it is low because the year-ago quarter carried a tariff overlay that has not repeated – not because reserves were released, which is a different thing and would carry a different signal. The impaired line, meanwhile, was up 6% year on year and down 16% sequentially, and the sequential direction is the newer information. The question is whether both hold: a performing provision that stays near zero while the impaired line keeps improving would confirm the cycle the market has been pricing, and either one turning would not. Watch the two lines separately rather than the total; the 34 basis-point impaired ratio against 40 in the prior quarter, which is the cleaner read than the 35 basis-point headline; net interest margin and volume growth into a labour market that has strengthened for three consecutive months and a resale market where prices are down but listings are down further; and Capital Markets, up 23% last quarter and the most cyclical of the five segments, alongside a 13.5% CET1 ratio that leaves room for buybacks. The read-across is immediate – Bank of Montreal and Scotiabank report two days earlier, so RBC either confirms what they describe or contradicts it – and the trade backdrop that drove last year’s overlay remains the obvious source of a reversal.
🇺🇸 CrowdStrike
$CRWD reports fiscal second-quarter results Wednesday after the US close, the same evening as Nvidia and Salesforce – a high-multiple software stress test scheduled into the busiest hour of the week. The setup changed sharply in the run-up. The shares closed Friday at $191.95, up 0.85%, but that is about 15% below the $225.53 close of 13 August and roughly 16% under a 52-week high of $227.50; the 52-week low is $85.68. Two things overlapped in that decline: a broad move out of premium security and software names, and the 20 August departure of chief technology officer Elia Zaitsev after more than thirteen years, to start an AI venture fund, with no successor named. On trailing revenue the stock still trades near 38 times sales, and around 175 times forward non-GAAP earnings on the multiples being quoted around the print.
The April quarter was strong. Revenue was $1.39bn, up 26%; ending annual recurring revenue reached a record $5.51bn; net new ARR was $256m, up 32% year on year; and the company posted record operating cash flow of $591m and record free cash flow of $468m. The Falcon platform now carries 33 modules, up from 30 a year earlier. Management raised full-year net new ARR growth guidance by 520 basis points, to about 27.7% at the midpoint, and announced a four-for-one stock split. Guidance for this quarter is $1.436–1.442bn of revenue and ending ARR of $5.790–5.795bn, up about 24%; management translated that range on the call into net new ARR of $284–286m, or 28–29% growth. Consensus sits close to the guide, near $1.44bn of revenue and $0.29 of post-split earnings. At least one large broker has put the practical bar higher, arguing that a net new ARR figure above roughly $292m – a beat of more than 3% – is what the valuation implies.
Our readNet new ARR is the swing factor, though it is worth being precise about what was actually guided: the company gave an ending-ARR range, and $284–286m is the net new ARR that range implies. At 28–29% growth that is a step down from the 32% delivered in April, but it sits in line with the roughly 27.7% the full year is guided to, so simply meeting it would not on its own say the trend has broken. Watch the mix behind the number, specifically how much comes from Falcon Flex expansion within existing accounts versus new logos, because consolidation-driven ARR and greenfield ARR do not carry the same durability; the full-year 27.7% net new ARR growth guide, which management raised only one quarter ago and now has to stand behind; and any commentary on the chief technology officer succession, which matters more than usual for a company whose case rests on shipping AI-native detection faster than its competitors. The drawdown cuts both ways: roughly 15% has already come out of the price in six sessions, so some of the perfection the multiple embedded is no longer there. The read-across runs to the premium-security cohort and, jointly with Salesforce a few minutes later, to whether AI-agent demand is converting into contracted revenue rather than pilots.
Get weekly earnings previews
Join the waitlist for early access to FinRyse – and receive our weekly market briefings.
🇺🇸 Salesforce
$CRM reports fiscal second-quarter results Wednesday after the US close, with the call at 5pm Eastern – the cleanest available test of whether AI is adding to enterprise software spending or redirecting it. The shares closed Friday at $209.17, up 1.82% on the day and about 9% above their 17 August close, but still roughly 22% below a 52-week high of $269.11 and down about 21% since the end of 2025; the 52-week range is $146.32 to $269.11 and the stock trades near 24 times trailing earnings. The de-rating has been the story of the year: analysts at KeyBanc flagged negative customer feedback on Agentforce in a chief-information-officer survey, and a July downgrade at Morgan Stanley cut the price target to $185 from $287 on the argument that Agentforce’s headline metrics were not showing up in contracted revenue.
The April quarter gave both sides something. Revenue was $11.13bn, up 13% and 12% in constant currency, including $444m from Informatica; subscription and support revenue rose 14%. GAAP operating margin was 21.1% and non-GAAP 34.8%, with non-GAAP earnings of $3.88, up 50%, and free cash flow of $6.6bn. Agentforce annual recurring revenue reached $1.2bn, up 205%, and Data 360 plus AI ARR reached $3.4bn, up 200%, of which $1.1bn was Informatica Cloud. Against that, current remaining performance obligation – the contracted book due within a year – was $33.6bn, up 14% and 13% in constant currency, and total RPO $67.9bn, up 11%. Guidance for this quarter is $11.27–11.35bn of revenue, 10–11% growth, with non-GAAP earnings of $3.25–3.27 and cRPO growth around 14%; consensus is near $11.33bn and $3.28. Informatica supplies slightly more than four percentage points of that 10–11%, which leaves organic growth somewhere in the mid single digits. The full year is guided to $45.9–46.2bn, with about three points of it acquired, and operating and free cash flow growing only 4–5% – halved from the 9–10% guided in February, and attributed by the company to the debt raised to fund a $25bn accelerated share repurchase rather than to anything in the operating business. Chief financial and operating officer Robin Washington said in May that the company remained confident in delivering organic revenue acceleration in the second half of the fiscal year.
Our readcRPO is the swing factor, not Agentforce ARR. Agentforce ARR is an annualised run rate rather than booked revenue, so the $1.2bn is not strictly comparable with the $45.9–46.2bn the company expects to report – but the order of magnitude is the point: something growing 205% off roughly two to three percent of the base can be entirely real and still leave group growth at 11%. That is exactly why the market has stopped treating the first number as evidence about the second. cRPO is where new contracted business has to appear, and the second-half organic acceleration management promised in May either becomes visible in it this week or does not. Watch cRPO against the roughly 14% guide; the split between acquired and organic, since slightly more than four points of the quarter’s reported growth is Informatica and mid-single-digit organic is the number the acceleration case has to be held against; the full-year revenue range, already set at $45.9–46.2bn; and any reconciliation of the 4–5% cash-flow guide with the buyback debt behind it. The bar here is durability rather than upside – the stock has already given back a fifth of its value this year – and the read-across runs to Workday the following evening on a closely related question, and to the wider claim that agentic AI is additive to seat-based software rather than a substitute for it.
🇨🇳 PDD Holdings
PDD opens the week, reporting June-quarter results Monday before the US open with a call at 7:30am Eastern – the first read on the Chinese consumer this week and the tone-setter for a Friday stacked with Chinese reporters. The shares closed Friday at $88.38, down 1.27%, within a 52-week range of $71.94 to $139.41; that is about 37% below the high and roughly 22% lower since the end of 2025. On trailing earnings the stock trades near 8.6 times, with a trailing net margin above 21% – a combination that tells you the market is discounting the durability of those margins rather than their current level. Consensus for the quarter is near $17.1bn of revenue and $2.77 of earnings, though the company has a record of printing revenue below the sell-side number and discloses less operating detail than its peers.
The March quarter is the base, and it repays a careful read. Total revenue was RMB 106.2bn, up 11%, with transaction services up from RMB 47.0bn to RMB 56.3bn – a line that carries Temu but also domestic transaction services, so it is not a clean proxy for either. Operating profit rose 22% to RMB 19.6bn and non-GAAP operating profit 15% to RMB 21.1bn, while net income fell 15% to RMB 12.5bn. That divergence sits almost entirely below the operating line: interest and investment income swung from a gain of RMB 223m to a loss of RMB 632m, and other income of RMB 3.26bn a year earlier became a loss of RMB 2.03bn. The investment spend shows up higher up the statement. Cost of revenue rose 15% to RMB 46.9bn on fulfilment, bandwidth and payment processing, taking gross margin down roughly 1.3 points to about 55.9%, while sales and marketing was held almost flat at RMB 33.8bn – which is why operating margin still improved, from about 16.8% to 18.4%. Research and development rose 24% to RMB 4.4bn. Cash and short-term investments stood at RMB 436.1bn. Behind those numbers sit two structural changes: a commitment of roughly RMB 100bn over three years to merchant support, supply-chain upgrades and brand development, and the loss of the de minimis exemption – suspended for China and Hong Kong in May 2025 and for all origins that August – which forced Temu to rebuild its US business around locally held, duty-paid inventory and domestically based sellers rather than direct parcel shipment.
Our readOperating margin is the swing factor, not the gap between operating profit and net income. That gap was mostly interest, investment and other items, which say little about how the business is being run; what the March quarter actually showed was gross margin giving way to fulfilment and supply-chain cost while marketing spend was held flat, and the open question is whether that trade can be repeated. Watch cost of revenue and the gross margin against the roughly 55.9% just printed; sales and marketing, unchanged year on year and the easiest line to flex in either direction, against a three-year RMB 100bn commitment that is a cost long before it is a moat; research and development, up 24%; and any disclosure on the US business after the pivot to local fulfilment, because a duty-paid, locally stocked model carries more working capital and, in principle, thinner unit margins than the parcel model it replaced. The read-across runs to cross-border e-commerce and to tariff-exposed retail more broadly – and the usual caveat applies more here than elsewhere: PDD publishes no standalone Temu numbers, so the margin trajectory will tell you more than any beat or miss against a consensus the company does not guide toward.
🇨🇳 BYD
BYD’s board meets Friday 28 August to approve interim results for the six months to 30 June, with the release expected after the Hong Kong close – the most consequential non-US industrial print of the week, and the first half in which the export business is large enough to change the shape of the group’s accounts. The H-shares closed Friday at HK$93.15, up 1.47% and about 5.6% higher over five sessions, within a 52-week range of HK$71.40 to HK$120.80 – roughly 23% below the high and about 16% lower over twelve months. The Shenzhen line closed at RMB 90.47.
The volume data is already public and it points two ways at once. First-half sales were 1,808,511 new-energy vehicles, down 15.7% year on year, with overseas sales up 70.7% to 792,256 units – about 44% of the global total – and domestic sales down close to 40%. July was the strongest month of the year at 419,211 units, up 21.8%, including a record 180,538 new-energy vehicles exported on the company’s own count, more than double the year-ago figure. Through seven months the total is 2,227,722 units, down 10.5%, against a full-year target of 5.0–5.5 million that has not been revised; reaching even the low end would require about 554,000 units a month from August onward, above the all-time monthly record of 514,809 BYD set in December 2024. The profit picture going in is weak: the March quarter brought revenue of RMB 150.2bn, down 11.8%, and net profit of RMB 4.08bn, down 55% and the lowest in more than three years, a fourth consecutive quarterly decline. The year-ago half is the yardstick – revenue of RMB 371.3bn, up 23.3%, net profit of RMB 15.5bn, up 13.8%, and research and development spending of RMB 30.9bn, itself up 53%. Full-year 2025 revenue passed RMB 800bn for the first time while net profit fell 19%.
Our readMix is the swing factor, and this is the first half where it is large enough to matter to the P&L. Roughly 44% of volume came from exports, which carry higher prices than domestic sales into a price war. Group revenue is the wrong place to test that, though: BYD also books electronics, batteries and other businesses, so revenue divided by vehicles is a number about the conglomerate rather than the car company. The automotive and related-products segment and its margin are the lines that answer the question. Watch that segment inside the group’s RMB 371.3bn year-ago half; gross margin, which the company has not disclosed for the period and which is where the domestic discounting actually lands; net profit against RMB 15.5bn a year ago, given the March quarter alone fell 55%; and research and development, RMB 30.9bn in the year-ago half and growing faster than revenue, now among the larger drags on reported profit. Higher export mix does not mechanically mean higher margin either – tariffs, localisation and the cost of ramping overseas plants all sit against the price premium. The full-year framing may matter more than the half: at 2,227,722 units through July, the 5.0–5.5 million target implies a monthly run rate the company has never achieved, so any restatement of it is arguably the more consequential disclosure. The read-across runs to Tesla’s China position, to European volume manufacturers meeting the export wave, and to the battery chain. It is also a Friday-evening Hong Kong release landing the same day as Jackson Hole, which limits how much of it gets traded in US hours.
Weekly Calendar
Some notable names reporting this week. “Post-close (exp.)” marks a confirmed reporting date with no stated publication time, where the timing follows the company’s established Hong Kong pattern rather than its own notice.
| Company | Country | Sector | Timing |
|---|---|---|---|
| Monday 24-Aug | |||
| PDD Holdings | 🇨🇳 | Discount & cross-border e-commerce | BMO |
| Tuesday 25-Aug | |||
| Bank of Montreal | 🇨🇦 | Diversified banking | BMO |
| The Bank of Nova Scotia | 🇨🇦 | Diversified banking | BMO |
| Intuit | 🇺🇸 | Tax & small-business software | AMC |
| Wednesday 26-Aug | |||
| NVIDIA | 🇺🇸 | AI accelerators & networking | AMC |
| CrowdStrike | 🇺🇸 | Cloud & endpoint security | AMC |
| Salesforce | 🇺🇸 | Enterprise software | AMC |
| CNOOC | 🇨🇳 | Offshore oil & gas | Post-close (exp.) |
| Synopsys | 🇺🇸 | Chip design software | AMC |
| Thursday 27-Aug | |||
| Royal Bank of Canada | 🇨🇦 | Diversified banking | BMO |
| Marvell Technology | 🇺🇸 | Custom AI silicon & networking | AMC |
| The Toronto-Dominion Bank | 🇨🇦 | Diversified banking | BMO |
| China Life Insurance | 🇨🇳 | Life insurance | Post-close (exp.) |
| Canadian Imperial Bank of Commerce | 🇨🇦 | Diversified banking | BMO |
| Autodesk | 🇺🇸 | Design & engineering software | AMC |
| Workday | 🇺🇸 | HR & finance software | AMC |
| Friday 28-Aug | |||
| Industrial and Commercial Bank of China | 🇨🇳 | Diversified banking | Post-close (exp.) |
| Agricultural Bank of China | 🇨🇳 | Diversified banking | Post-close (exp.) |
| China Construction Bank | 🇨🇳 | Diversified banking | Post-close (exp.) |
| PetroChina | 🇨🇳 | Integrated oil & gas | Post-close (exp.) |
| Bank of China | 🇨🇳 | Diversified banking | Post-close (exp.) |
| China Shenhua Energy | 🇨🇳 | Coal mining & power | Post-close (exp.) |
| BYD | 🇨🇳 | Electric vehicles & batteries | Post-close (exp.) |
| Meituan | 🇨🇳 | Local services & delivery | AMC |