Web Research
Web Research
The Bottom Line from the Web
The web layer is empty for this run. Every external research phase — Industry, Warren, Quant, Sherlock, Historian, Forensic, and the subsequent specialist-query phase — failed at the provider call with HTTP 402 (insufficient credit) on 2026-05-27. No third-party search results, news articles, analyst notes, or page texts were retrieved. The investment thesis on this tab therefore relies entirely on the 20-F, transcripts, financial statements, and governance disclosures already covered by the other specialists; the open questions that the web would have answered are catalogued below as a follow-up agenda.
Provider failure on 2026-05-27 10:38 UTC. data/web-research/research-preload.json shows status: skipped, provider_status: unavailable, reason: parallel_insufficient_credit. The same 402 error blocked all six preload phases and the downstream specialist-query phase (specialist-research.json, exit_code 1, 0 unique queries fetched, 0 pages read). This is not an editorial choice — it is a hard infrastructure gap.
What Matters Most
The five items below are the highest-impact questions the web should have answered. They are ranked by how much each answer would shift the central bull/bear math (cash-fortress + shrinking operating stub + Haier-controlled outcome).
1. Haier's strategic plan for ATHM is the single largest swing factor and remains a black box
Haier's CARTECH paid US$1.8B for 43.6% in August 2025 (an ~88% premium to ATHM's market cap), replaced five of eleven directors, and installed Chi Liu — a career home-appliance executive — as CEO/Chairman. The 20-F discloses the transaction terms but not the strategic rationale, post-deal integration plan, related-party flows with Haier affiliates, or capital-allocation intent (special dividend, take-private, operational reset). Every other specialist flagged this as the dominant variable; the web phase that would have surfaced Chinese-language press coverage, dealer-channel commentary, and CARTECH disclosures was the one that failed.
Implication: Until the FY2026 20-F lands or Haier issues a strategic statement, every bull/bear scenario on ATHM is conditional on a controller intent that no public source has disclosed.
2. Dongchedi's actual DAU and pre-IPO financials remain unknown
Dongchedi (ByteDance) is the threat that the Moat and Competition specialists rank as the single most important competitive risk. The internal analysis assumes ATHM holds a ~2× DAU lead based on QuestMobile MAU comparisons (ATHM 77.5M mobile DAU Dec 2025 vs Dongchedi ~35.7M MAU mid-2024). A Hong Kong IPO targeting ~US$1.5B is reportedly planned for 2026 — at which point Dongchedi must disclose dealer count, ARPD, ad take-rates, and segment revenue. None of that prospectus-grade data exists yet in this run because the web phase failed to fetch any 2025-2026 references.
Implication: The narrow-moat conclusion rests on a competitive gap that has not been quantified with hard daily-active or ad-revenue numbers in over 18 months.
3. The 27 February 2026 volume distribution day has no identified catalyst
The Tech specialist flagged a 14× volume distribution day on 27 February 2026 as the largest on record and asked the web to attribute it — a dealer-subscription cut, Ping An residual block trade, ADR-cohort regulatory news, or earnings-related disclosure. None of these candidates were on-file in this run, and the search that would have closed the attribution gap returned no results.
Implication: A material price/volume event in the last 90 days cannot be tied to a specific cause from internal data alone.
4. No external confirmation of the goodwill, HNTE renewal, or short-seller history
Forensic asked the web to verify three independent items: (a) any SEC comment letter on the unchanged RMB 3,942M / $563M goodwill carried flat since FY2021 despite a 76% drop in operating income from the FY2020 peak; (b) whether Autohome WFOE and Beijing Autohome Technologies have renewed their High and New Technology Enterprise (HNTE) status for the 2026-2028 cycle (failure would lift ETR from ~9% toward 15-25%); and (c) any prior short-seller report on the VIE structure, revenue recognition, or TTP acquisition pricing. All three remain unverified externally.
Implication: The "ICFR effective, no restatement" baseline is intact from filings, but three discrete external confirmations that would harden the forensic verdict are missing.
5. Sell-side FY26/FY27 FCF dispersion and any post-Q4 FY25 revisions are not captured
Quant's scenario analysis hinges on a base case FY26 FCF of $90-180M. The internal analyst_estimates.json carries revenue and EPS dispersion but not FCF or EBITDA, and there is no captured post-Q4-FY25 revision flow (upgrades, downgrades, target-price changes, or coverage drops). The web research phase that normally pulls broker reports and consensus revisions did not run.
Implication: Position sizing against the FCF watch metric is currently anchored to internal modelling, not to a live consensus comparison.
Recent News Timeline
The intended news timeline is empty. Industry specialist's manifest does record that 6 direct WebSearch calls (outside the Parallel pipeline) were attempted; those results were folded into the Industry tab, not into a structured news feed. For a current news scan, an analyst should run WebSearch manually on Bloomberg, Reuters, Caixin (Chinese-language), and 21st Century Business Herald (21cbh.com) covering 2026-02-01 through 2026-05-27.
What the Specialists Asked
Each tab below contains the questions the specialist authored for the web phase, the reason it matters (i.e., what hinges on the answer), and an honest answer state: Unanswered — Parallel preload failed for all of them. The questions are the deliverable — they form the highest-quality external-research backlog for ATHM coming out of this deep-dive run.
Governance and People Signals
The web phase would normally surface ISS/Glass Lewis recommendations on the August 2025 board reshuffle, Chinese-language coverage of Chi Liu's CARTech track record, and any proxy-fight or activist commentary. None of that was retrieved here. The signals available from internal sources are summarised — every line below is from filings, not from the web.
Governance external confirmations still missing: (a) CARTECH purchase price and minority-shareholder protections from the share-purchase agreement; (b) Chi Liu's prior CARTech operating track record; (c) ISS/Glass Lewis recommendations on the August 2025 board changes; (d) any short-seller campaign history. All four are in Sherlock's web backlog.
Industry Context
The Industry specialist already published the structural primer (mature China online auto vertical, low industry attractiveness, high regulatory risk, NEV penetration at 60.4% retail in December 2025, ~4,000 4S closures forecast for 2025, 55.7% of 4S dealers loss-making). That primer used six direct WebSearch calls — the only web data that successfully reached the run. No new or thesis-changing external evidence was added in this web-research phase that is not already captured in the Industry tab.
The structural external picture that would materially update the industry view but is currently absent:
Dongchedi prospectus. Whenever ByteDance files the Hong Kong listing prospectus reportedly targeted at ~US$1.5B for 2026, the disclosed dealer count, ARPD, ad take-rate, and segment revenue become the cleanest competitive benchmark that exists. That single document would re-rate the entire China online-auto comp set and is the most consequential external industry event on the watchlist for this name.
OEM-direct app share-of-attention. A QuestMobile or similar third-party measurement of BYD / NIO / Li Auto / Xiaomi / Xpeng consumer-app MAUs against vertical-platform MAUs would tie the OEM-direct threat severity to a measurable share-of-attention number rather than just NEV penetration trajectory.
Vertical-platform ad-spend share. The split of OEM auto-advertising budgets across vertical platforms vs Douyin / short-video vs OEM-direct apps vs traditional media is the cleanest single industry-wide test of whether vertical platforms have lost the toll-road position. Currently inferred from media-revenue trends only.
Method Note
This page was produced from data/web-research/research-preload.json, data/web-research/specialist-research.json, and every specialist's *-claude-queries.json file. The query files are the deliverable when the external provider is offline — they enumerate the questions that an analyst should run manually before the next full deep-dive on this name. The next time the Parallel provider is available, re-running the preload + specialist phases will populate this tab end-to-end with cited findings against those exact questions.