Dolphin Research

0519 | Dolphin Research Focus: 🐬 Macro/Industry 1) Trump announced on social media that, at the request of leaders of Qatar, Saudi Arabia and the UAE, the planned strike on Iran set for the 19th has been postponed. He said the US and Iran are in 'serious talks', while the three Gulf states fear an escalation could hit their own energy infrastructure. He also stressed that US forces remain on alert, warning that if talks collapse, a full-scale offensive would follow.

Below is Dolphin Research's curated Trans of $Baidu(BIDU.US) FQ1 2026 earnings call. For our earnings read-through, please see 'BIDU: Little Else to Rely On, All-in on Kunlun Chip'.

Key takeaways from the results.

1) Shareholder returns: last quarter the company announced a new buyback and introduced its first dividend policy. Management reiterated it will balance long-term AI investment with shareholder returns, committing to deliver sustainable value for shareholders.

2) AI core revenue exceeded 50% of total for the first time. In Q1 2026, BIDU's AI core revenue came in at over RMB 13.6bn...

$Baidu(BIDU.US) Q1 results were a mixed bag, with sharp contrasts across lines. They reflect a meaningful strategic pivot and business realignment in recent quarters.

By contrast, the structurally weak ads business is no longer the focus. What matters to investors now is execution on the narratives BIDU can sell—AI Cloud, Kunlun chips and buybacks.

Last quarter, the company adopted a new reporting framework. Results are split into 'AI' and 'Legacy' businesses...

BIDU 1Q26 First Take: Q1 was mixed, a tale of two halves, reflecting recent strategic shifts and a business pivot. With legacy ads already structurally weak, investors seem more focused on whether AI Cloud, Kunlun Chip and buybacks are tracking to plan.

As most sell-side models still follow the old framework, Dolphin Research blends the old and new to surface expectation gaps. This also helps show growth in BIDU's new AI lines.

1) Highlights first: the pillars delivered.

(1) AI Cloud beat. Q1 revenue was RMB 8.8bn (+79% YoY), vs. sell-side at RMB 7.0–8.0bn; GPU compute revenue rose 184% YoY.

(2) Opex dropped sharply, margins improved. Notably, opex as a share of revenue fell to 31% from 45% in Q4, largely on lower personnel costs.

SBC tells the same story: stock-based comp fell 24% YoY in Q1. With a higher market cap vs. last year, this implies a sizable reduction in headcount eligible for awards.

Using the old framework, Dolphin Research roughly split margins for ads and cloud. Trend-wise, cloud margins appear to have improved.

2) Pain points: are AI apps losing steam?

(1) Legacy ads -28%. This was expected and there is little to add. Holiday timing played a role, but macro pressure and strategic shifts reflected in headcount changes suggest no inflection yet.

(2) AI apps and AI-native ads are losing momentum. Sequentially they fell 7% and 12%, and AI app revenue (incl. Baidu Wenku, autonomous driving, cloud drive and digital employees) slowed to flat YoY, which is out of sync with industry strength. We flagged this last quarter, likely driven by competition.

3) Buybacks started, likely constrained by CNY and blackout. The two-year $5bn program announced in Feb saw $172mn repurchased in Q1 (roughly one month), which looks a bit light on pacing; we will watch the call for plans ahead.

That said, management appears more committed to shareholder returns this time. With net cash of ~RMB 100bn (ex short-term borrowings) and ample FCF, we expect execution as planned.

4) Investment stepping up. Q1 capex nearly doubled YoY to RMB 5.8bn. Still low vs. other model/cloud majors, and AI Infra revenue of RMB 8.8bn alone more than covers capex.

5) Watch for commentary on Kunlun Chip's IPO plan on the call. In early May, Kunlun Chip completed IPO coaching registration; on a typical timeline, listing could be completed by end-Q3. Given the early filing (submitted in early Jan), a mid-to-early Q3 listing is also possible.$Baidu(BIDU.US)$BIDU-SW(09888.HK)

As Dolphin Research had warned, consumption pressure in Q2 is significant, and April data confirmed a clear slowdown. Retail sales came in notably weaker.

0518 | Dolphin Research Focus: Macro/Industry. NBS data show Jan–Apr retail sales at RMB 16.49tn (+1.9% YoY).

Below are Dolphin Research's notes from H World's FY26 Q1 earnings call.

For our full take on the results, please see 'H World: Shedding Burdens, Going Lean—Still a Top Student in Hotels!'

RevPAR up further YoY

Below is Dolphin Research's Trans of $LEAPMOTOR(09863.HK) Q1 26 earnings call. For our earnings take, see 'Leapmotor: Aiming for 1mn, is GPM slipping first?'. I. Key highlights: 1) Guidance — Q2 26 delivery guidance of 240–250k units (April alone exceeded 70k)...

$LEAPMOTOR(09863.HK) released its Q1 2026 results after the HK close on May 15 (BJT). The print marked a departure from Leapmotor’s historically steady execution.

Despite a seasonal soft quarter for the industry, volumes were still decent and opex discipline was reasonable. However, both revenue and GP missed, with per-vehicle GP well below street expectations.

On the details: revenue missed as ASP fell further on a shift toward lower-end models. Q1 revenue was RMB 10.82bn (+8% YoY)...

Huazhu 1Q26 First Take: Overall revenue held up well in Q1, and operating metrics kept recovering after turning positive in H2 last year.

1) Revenue growth ticked up QoQ. Total revenue was RMB 6.0bn, up 11% YoY, with a modest sequential acceleration vs. Q4.

Franchise revenue rose 20% YoY to RMB 3.0bn, driven by more rooms and RevPAR back in positive territory, lifting single-hotel GMV.

2) RevPAR expanded further YoY. RevPAR rose 2.9% YoY to RMB 214 per night.

3) Solid openings with higher-quality growth focus. Net adds were 357 hotels in Q1 (537 openings, 180 closures), keeping a fast pace.

4) Step-up in sales spend: With a higher franchise mix YoY, GPM expanded 580bps to 39%.

Leapmotor 1Q26 First Take: Overall, the first quarter of FY26 did not carry forward the company's usual operational discipline. The print is concerning.

Despite a decent delivery print for an industry off-season and reasonable cost control, both revenue and GP fell short. Revenue was RMB 10.82bn, a slight miss vs. market. GP missed by about 40%, which weighed on sentiment.

On margins, GPM fell 550bps YoY; even stripping out roughly RMB 200mn of GP uplift from strategic partnerships, GPM was still down 390bps YoY. Management added that lower output increased per-unit manufacturing costs.

The market has largely priced in several factors.

1) Management had guided to a sequential decline for Q1 due to seasonality on the Q4 call. This was flagged in advance.

2) Mix shifted down-market as T03 rose from 11% in Q4 to 22% in Q1, while the lower-price A10 started to ramp in Mar, pressuring ASP. This further diluted the model mix.

On opex: selling expenses were RMB 680mn, seasonally softer and broadly in line. R&D was RMB 1.04bn, a slight miss tied to project timing. G&A came in at RMB 440mn vs. the market’s RMB 420mn, essentially in line and largely seasonal.

OP was -RMB 410mn, better than the market’s -RMB 540mn, implying sizable non-recurring items this quarter. More color should come on the call.

a) Management’s explanation for the per-vehicle margin decline. Details will help quantify the impact.

b) Q2 guidance. It will frame the near-term recovery trajectory.

c) Sustainability of the strategic partnership uplift. This affects visibility on GP support. Dolphin Research will publish a follow-up note; stay tuned. $LEAPMOTOR(09863.HK)

0515 | Dolphin Research Focus: 🐬 Macro/Industry 1) Today is Jerome Powell's final day as Fed Chair, and Kevin Warsh begins his first term. On the same day, Governor Stephen Miran tendered his resignation and will step down as Warsh takes office.

The following is Dolphin Research's transcript of AMAT FY26 Q2 earnings call. For our takeaways, see 'Capex Boom Underway: Can AMAT Capture the Windfall?'.

The following is Dolphin Research's transcript of SMIC's FY26Q1 earnings call; for our take, see 'Price hikes on one hand, new narratives on the other — is SMIC about to surge?'. I. $SMIC(00981.HK) results highlights.

1) Shareholder returns: No payout for 2025 as FCF remains negative. The company is in a critical capacity expansion phase, prioritizing capacity build-out and R&D.

2) Q2 guidance: revenue up 14%-16% QoQ on stronger shipments and ASP. GPM at 20%-22%...

Applied Materials (AMAT) posted its FY26 Q2 results (quarter ended Apr 2026) after hours early May 15 (Beijing time), with the following highlights: 1) Core metrics: revenue of $7.9bn (+11% YoY), beating the Street at $7.7bn, driven by AI compute infrastructure build-out that lifted demand for advanced logic, DRAM, and advanced packaging tools. $Applied Materials(AMAT.US) GPM was 49.9%, up 90bps QoQ...

AMAT First Take: Revenue and GPM were solid, both beating estimates, as growth was driven by the AI compute buildout, which lifted demand for advanced logic, DRAM and advanced packaging tools.

Guidance for next quarter looks even stronger than the print. Mgmt guides revenue of $8.45-9.45bn, midpoint +13% QoQ, well above the Street at $8.15bn. EPS is guided at $3.16-3.56, also ahead of the $2.68 consensus.

As downstream customers raise capex, AMAT is set to enter a revenue expansion phase. Demand is particularly strong for tools in advanced logic, memory and advanced packaging.

On the other hand, new product iterations should lift GPM and further improve operating metrics. After the blowout guide, the stock rose as much as 7% after hours.

Breaking down the next-quarter guide, semi equipment growth is already near ~30% in the upcoming quarter. In other words, with capex still rising, the full-year outlook implies limited acceleration thereafter, which poured cold water on sentiment.

Overall, after a sustained rally, mgmt's conservative full-year stance cooled near-term enthusiasm. Still, upstream semi equipment remains a high-conviction leg in this upcycle. As chip pricing and capacity additions pick up, fundamentals should inflect from a clear bottom for $Applied Materials(AMAT.US).

SMIC (0981.HK/688981.SH) posted its Q1 2026 results after the HK close on May 14 (Beijing time). The quarter covered the period ended Mar 2026. Key takeaways: 1) Headline results:

$SMIC(00981.HK) reported revenue of $2.5bn, slightly below market est. ($2.52bn). Revenue rose 0.7% QoQ, broadly in line with guidance (flattish QoQ). Growth in industrial and auto end-markets was largely offset by weakness in smartphones...

SMIC 1Q26 First Take: Results were solid. Revenue was roughly in line with the Street, while GPM beat meaningfully.

On margin drivers, ASP rose by approx. $9/wafer QoQ, while unit cost fell by $2/wafer, lifting GPM this quarter. Recent demand recovery in power-related products is driving price increases at mature nodes, which should support further margin expansion.

Versus the print, next quarter guidance matters more, and management expects it to be 'quite good'.

The company guides revenue up 14%–16% QoQ, well ahead of the Street at +7%. GPM is guided to 20%–22%, with the midpoint above the Street's 20.5%.

For next quarter, Dolphin Research sees both volume and price lifting results. Demand recovery at mature nodes should support pricing, while ongoing capacity additions and higher utilization should also help.

While margins and earnings remain at a relatively low base, guidance suggests the company is turning up from the trough.

Among second-tier foundries, SMIC still carries expectations for China's push into advanced nodes. That has been the key reason it enjoyed a valuation premium in the past.

However, with recent price hikes in some mature-node categories, rallies in UMC and GlobalFoundries have largely erased SMIC (H)'s relative valuation advantage.

SMIC's recent share-price underperformance vs. peers stems from two factors: first, power and other mature-node products account for a relatively small mix; second, the market worries about a MATCH Act-driven escalation of sanctions on immersion DUV. Dolphin Research remains relatively confident in 2H growth. Smartphone demand should bottom in Q2 (per Qualcomm management), and the Ascend 950DT is moving into mass production.

Price increases now seen in some mature-node categories could broaden into an upcycle for the wider semiconductor industry. With demand from Chinese AI customers (e.g., Cambricon), as long as advanced nodes can ship smoothly, valuation upside remains for the company.$SMIC(00981.HK) $SMIC(688981.SH)

Below is Dolphin Research's recap of $JD.com(JD.US) FY26 Q1 earnings call.

I. Core financial highlights review

$JD.com(JD.US) released its Q1 2026 results on May 12. Takeaway: growth has bottomed and is recovering, with profits ahead of expectations. Total revenue was RMB 315.7bn (+~5% YoY), improving from +1.5% in the prior quarter and slightly above market expectations.

0514 | Dolphin Research Focus: 🐬 Macro/Industry

Below is Dolphin Research's summary of BABA's FY26 Q4 earnings call transcript.

$Alibaba(BABA.US) released its last-quarter FY26 results pre-mkt on May 13. As widely expected, headline performance was weak this quarter.

Below is Dolphin Research's Trans of $TENCENT(00700.HK) FY26 Q1 earnings call. For our post-results take, see 'Tencent: No Longer Resting on Past Laurels; AI Is the Way Forward'.

1) Shareholder returns: Q1 buybacks totaled approx. RMB 7.9bn. Management believes the stock is undervalued, is accelerating monetization of the investment portfolio to fund continued buybacks through the year, and sees now as a particularly attractive window for repurchases.

2) Outlook: CapEx is expected to increase materially vs. last year. Guidance points to a sharp step-up this year...