← Back to All Research
Commission a Report
← Back to All Research
Commission a Report
The September 2025 report named four companies as "AI-era winners." Six months later the portfolio's track record is deeply uneven: one strong call, one ambiguous outcome, and two picks that must be called analytical failures.
| Ticker | Verdict | Return (est.) | One-line Summary |
|---|---|---|---|
| GOOGL | STRONG WIN | +78% | Antitrust relief landed the day after publication; AI thesis fully confirmed |
| TSLA | MIXED | +31% | Robotaxi milestone real, but vehicle business in accelerating decline |
| TDOC | THESIS BROKEN | -37% | Structural decline continues; AI framing was wishful, not analytical |
| CHGG | ANALYTICAL FAILURE | -74% | Sub-$1 stock, NYSE delisting notice, revenue down 49% YoY |
| Equal-weighted portfolio | Marginal | -0.5% | GOOGL rescues a deeply negative three-name outcome |
| S&P 500 (same period) | Benchmark | -3.0% | Index peaked late 2025, reversed in early 2026 |
Blended alpha vs. S&P 500: +2.5 percentage points — but this is entirely attributable to one name. Strip GOOGL and the three-name portfolio would have returned approximately -27%, underperforming the index by 24 points. That is a portfolio construction failure, not a research success.
| Ticker | Entry Price (Sept 4, 2025) | Current Price (Mar 25, 2026) | 6-Month Return | Alpha vs. S&P |
|---|---|---|---|---|
| GOOGL | ~$165 (est.) | ~$293 | +78% | +81 pp |
| TSLA | ~$290 (est.) | ~$380 | +31% | +34 pp |
| TDOC | ~$8.50 (est.) | ~$5.35 | -37% | -34 pp |
| CHGG | $1.91 | ~$0.50 | -74% | -71 pp |
| Portfolio | — | — | -0.5% | +2.5 pp |
| S&P 500 | ~5,700 | ~5,530 | -3.0% | — |
Entry prices estimated from contemporaneous trading data. GOOGL entry is pre-ruling close (antitrust ruling September 2; stock surged September 3-4, the day before and day of publication). TSLA September 2025 average close was $395.89; early-month entry is estimated lower. TDOC 52-week high ~$9.94; September 2025 context implies ~$8-9 range. CHGG September 12, 2025 price: $1.91 (sourced directly). S&P 500 return is price-return, not total return.
Alphabet was the defining AI-era infrastructure play: dominant search cash flows funding the Gemini AI buildout, Google Cloud compounding at scale, and an antitrust overhang that the market was dramatically mispricing. The core thesis was that if the DOJ's most severe remedies — forced divestiture of Chrome and Android — were avoided, the stock would re-rate sharply.
Antitrust ruling (September 2, 2025) — the day before publication:
Judge Amit Mehta declined to order the sale of Chrome or Android. Instead, he imposed behavioral remedies: a ban on exclusive search distribution contracts and a requirement to share certain search index data with competitors. Alphabet surged 8-9% in two sessions, adding over $230 billion in market cap.
Q4 2025 earnings (reported February 4, 2026):
2026 capex guidance: $175–185 billion for AI compute and Cloud — a statement of extraordinary demand conviction.
Stock trajectory:
Antitrust appeal: The DOJ and 35 states filed appeal on February 4, 2026, seeking structural breakup. D.C. Circuit expected to rule by late 2026. Market has repriced this as a long-duration tail risk, not an imminent event.
Separate ad-tech case: Judge Brinkema found Google monopolized publisher ad servers and ad exchanges (April 2025). DOJ wants Google to sell AdX. This is an ongoing overhang.
STRONG WIN. Thesis vindicated.
The antitrust relief catalyst played out almost exactly as anticipated. AI monetisation has moved from speculative to demonstrated. The DOJ appeal introduces future uncertainty but does not alter the near-term investment case.
Action: Maintain and strengthen coverage. Update price target. Track DOJ appeal timeline.
Tesla was framed as an AI-era winner on the strength of FSD monetisation, Robotaxi commercial launch, Optimus humanoid robots, and Musk's perceived ability to deploy AI at physical scale. The vehicle business was acknowledged as under pressure but treated as a bridge to the autonomous future.
The round-trip:
From an estimated ~$290 entry in early September 2025, TSLA peaked near $489 in December 2025 (+69%), then fell to the low $300s by late February 2026 as brand damage data, delivery disappointments, and political fatigue dominated. The stock has partially recovered to ~$380 as of late March 2026.
FY 2025 fundamentals — deteriorating:
Brand damage:
AI narrative catalysts — partial delivery:
2026 capex: ~$20 billion for new factories, Optimus production lines, and AI compute.
MIXED. Not a clean vindication.
The original thesis identified real optionality (robotaxi, Optimus) that is beginning to play out. But framing Tesla as an AI-era winner priced in a functioning vehicle business that does not exist. At +31% from estimated entry, the return is positive but the path was a violent round-trip driven by speculative sentiment, not fundamental improvement. Brand damage and delivery decline are structural, not cyclical.
Action: Maintain coverage with explicitly cautious framing. Robotaxi expansion in 2026 is the decisive variable. Musk political risk is non-trivial.
Teladoc was apparently framed as an AI beneficiary in digital health: AI-assisted virtual care improving unit economics, the chronic care platform (BetterHelp, primary care) offering durable revenue streams, and the company positioned to benefit from healthcare digitisation trends.
The thesis was incorrect. The data available in September 2025 should have raised larger red flags.
FY 2025 financial results:
2026 guidance — no recovery in sight:
Balance sheet — not the primary risk:
Analyst downgrades:
The structural problem: Teladoc is not an AI winner — it is an AI victim. Its virtual care product is being squeezed by AI-native mental health and primary care apps that operate at a fraction of Teladoc's cost. BetterHelp, Teladoc's direct-to-consumer therapy unit, is competing with AI therapy tools and free alternatives. The moat never existed or has been fully eroded.
THESIS BROKEN. This was the wrong call.
Teladoc should not have been framed as an AI-era winner in September 2025. It was already in structural decline, competing against AI-native products with superior economics and accelerating adoption. The AI narrative as applied to Teladoc was aspirational, not analytical.
Down approximately 37% from estimated entry.
Action: Remove from coverage. No credible re-entry thesis at current juncture. Consider monitoring for strategic transaction.
Chegg was apparently positioned as an AI turnaround story: a deeply discounted edtech platform (~$1.91 at time of report) that could leverage AI-assisted tutoring to stabilise its user base and pivot toward skills-based learning.
The deterioration from September 2025 to March 2026 is not a surprise outcome — it is a continuation of trends that were clearly visible at the time of the original report.
Chronology of collapse:
| Date | Event |
|---|---|
| Sept 4, 2025 | Original report published; CHGG ~$1.91 |
| Oct 26, 2025 | Chegg announces it will remain standalone (sale process failed) |
| Oct 27, 2025 | CEO Nathan Schultz replaced by returning founder Dan Rosensweig |
| Nov 2025 | Restructuring: 388 roles eliminated (~45% of global workforce) |
| Dec 12, 2025 | NYSE issues continued listing standard notice (30-day avg. below $1.00) |
| Feb 9, 2026 | Q4 2025 earnings: revenue $72.7M, down 49% YoY |
| Mar 2026 | Stock ~$0.50; company flagging potential reverse stock split |
Q4 2025 financials:
Q1 2026 guidance: Total revenue $60–62 million. Annualised run-rate under $250 million vs. peak revenues once exceeding $700 million.
The core problem: Chegg's subscription academic tutoring product is not being "disrupted" — it is being replaced. ChatGPT, Claude, Perplexity, and other AI tools answer the same homework questions for free, with equal or superior quality, available 24/7. Chegg had no technical moat, no proprietary data advantage, and no pricing power. This was visible in September 2025.
The "AI-era winner" label was precisely backwards for Chegg. Chegg is one of the clearest examples of an AI-era loser — a human-operated service platform directly displaced by large language models.
No active institutional Buy ratings. The stock trades below $1.00 and faces NYSE compliance requirements. Most institutional mandates cannot hold sub-$1 equities. The company is exploring a reverse stock split, which would be dilutive to existing holders and typically signals distress rather than recovery.
ANALYTICAL FAILURE. The thesis was wrong before the ink dried.
The disconfirming evidence was overwhelming in September 2025: revenue in freefall, subscribers abandoning the platform, AI competitors entrenched and improving, management under pressure, stock already down 95%+ from peak. No research process applying honest disconfirmation testing should have included Chegg in a list of AI-era winners.
Down approximately 74% from entry. NYSE delisting risk. No recovery catalyst for core academic product.
Action: Immediate removal from coverage. Acknowledge this as a research process failure in the internal record.
| Metric | Value |
|---|---|
| S&P 500 Sept 4, 2025 (approx.) | ~5,700 |
| S&P 500 March 25, 2026 (approx.) | ~5,530 |
| S&P 500 6-month return | -3.0% |
| PRZC portfolio blended return (equal-weighted) | -0.5% |
| Active alpha generated | +2.5 pp |
| Alpha ex-GOOGL (3-name portfolio) | -24 pp |
Context: The S&P 500's weak 6-month showing reflects a strong late-2025 market followed by a difficult early 2026. The index returned ~18% for full-year 2025 but gave back ground through Q1 2026 amid tariff uncertainty, elevated valuations, and macro volatility (YTD 2026: approximately -3% as of late March). This is a relatively flat environment in which individual stock selection should dominate.
The PRZC portfolio generated marginal alpha (+2.5 pp) in a flat market, but through a single winning bet. The concentration of outcome in GOOGL, combined with deep drawdowns in TDOC and CHGG, represents a poorly constructed portfolio regardless of the headline blended return.
GOOGL worked because the original thesis identified:
Lesson: Asymmetric catalysts with clear timelines are where high-conviction calls belong. The antitrust thesis was specific, falsifiable, and correct.
TDOC failure: The AI framing was applied in hope rather than analysis. A company with declining revenues, no demonstrated AI monetisation, and facing commoditisation from below is not an AI winner regardless of the press releases it issues. Future coverage of digital health names must separate "uses AI in marketing" from "benefits structurally from AI."
CHGG failure — more serious: The disconfirming evidence was not marginal or ambiguous. It was overwhelming. A stock that has fallen 95% from peak, a product directly substituted by free AI tools, revenues declining 40-50% quarterly — these are not signals to "buy the turnaround." The original inclusion of CHGG suggests either insufficient disconfirmation testing or a willingness to dress up a deeply troubled name as a growth story to complete a list. Neither is acceptable.
The four-name portfolio inadvertently mixed two fundamentally different categories:
Category 1 — AI infrastructure and platform builders (GOOGL, arguably TSLA):
Companies building or operating the AI stack. They benefit from AI capex, AI adoption, and AI monetisation directly.
Category 2 — AI-disrupted service incumbents (TDOC, CHGG):
Companies whose core product is a human-mediated service that AI tools are replacing or commoditising. Virtual tutoring. Virtual therapy. These are not AI winners; they are casualties.
Any future AI-era thesis must explicitly screen for which category applies. Conflating the two is the primary research failure of the September 2025 report.
Before naming a company as an "AI-era winner," PRZC Research analysts should be required to answer:
CHGG and TDOC would have failed questions 1, 2, and 3 in September 2025. The original report did not apply this filter.
| Ticker | Rationale |
|---|---|
| NVDA | Definitional AI infrastructure winner; may be over-covered but warrants position |
| AMZN | AWS AI monetisation + logistics automation; diversified |
| PLTR | Government and enterprise AI deployment; demonstrated revenue growth |
| NOW | ServiceNow; enterprise AI workflow automation with strong margins |
Entry price methodology:
S&P 500:
Current prices as of approximately March 24-25, 2026, sourced from multiple financial data providers.
PRZC Research | Investment Analysis Division | Scorecard prepared: March 25, 2026
Original report published: September 4, 2025
This document is for internal review purposes. It does not constitute investment advice. Past performance of research calls is disclosed for institutional accountability and process improvement.
Commission Research
Commission a bespoke PRZC report on any company, sector, or market question. Full CBOM methodology. Delivered within 5–7 business days. £500 per report.
Commission a Report — £500Bulk credits available from £350/report · Browse all free research