Tempus AI, currently one of the most polarizing stocks in the U.S. market.
On one side, Cathie Wood has listed it as the largest holding in the ARKG fund (9.49% weight), accumulating heavily at a price of $43. On the other side, there is the Spruce Point short report, collective lawsuits, the CEO selling over $34 million, and a short ratio of 18%.
On one side is the world's largest AI healthcare data platform. On the other side are the questions about "funds repatriation" and the whirlpool of related-party transactions.
Is it that Cathie Wood sees something others do not, or has she fallen into a Groupon-style trap again?
I spent a lot of time dissecting Tempus's SEC financial reports (10-K/8-K/S-1), the Spruce Point short report, legal documents, ARK's daily trading data, and cross-validating 10 valuation methods.
Conclusion: The data flywheel moat is unique, but management governance is the biggest X factor.
I. What is Tempus AI? In a sentence
It's the medical version of Palantir.
Palantir aggregates fragmented data from governments and businesses, using AI to create intelligence value. Tempus does the same thing—except its data sources are hospitals, and its clients are pharmaceutical companies and doctors.
Specifically, Tempus aggregates patient data from over 4,500 US hospitals—including genome sequencing results, electronic medical records, imaging reports, and pathology slides—and after cleaning and structuring all of it, it forms a database of over 40 million patient records and over 350 PB.
Then monetize from both ends:
For doctors, we provide AI-driven precision treatment matching (what is your tumor gene mutation → what is the most likely effective drug → which clinical trial is suitable for you).
We license de-identified, multimodal data to pharmaceutical companies to help them with drug discovery, clinical trial design, and patient recruitment. 19 of the top 20 pharmaceutical companies are our clients.
This is the "data flywheel": more testing → more data → better AI → attracting more doctors → more testing... At the same time, data is collected once and authorized to multiple pharmaceutical companies multiple times.
II. Why did this company catch my attention?
Three words: a sense of wrongful killing.
From $104 in October 2025 to the current $43, it has fallen by 59%. I have broken down the reasons for the pullback:
Panic over sharp drop in growth: FY2025 revenue grew by 83%, but FY2026 guidance is only 25%. The market was alarmed. However—about 50% of the 83% growth came from the $600 million acquisition of Ambry Genetics. Excluding the acquisition, organic growth is 28-33%, with Q4 organic growth at 33.5%. The 25% guidance is due to the high base effect of the acquisition, not a deterioration in business.
Short selling and litigation against Spruce Point: The stock plummeted 19% on the day the short-selling report was released in May 2025, subsequently triggering a securities class-action lawsuit. Allegations include related-party transactions inflating revenue, "fund repatriation" from the SoftBank joint venture, and aggressive billing practices by Ambry. How substantial are these allegations? More details later.
Insider selling spree: 118 insider sales and zero insider purchases in the past six months. CEO Lefkofsky is expected to reduce his holdings by over $34 million between November 2025 and January 2026. Although these are all part of the 10b5-1 pre-set plan, the signal is not good.
Tech stocks pullback: With a Beta as high as 5.22, tech stocks will fall even more as the broader market falls.
The question is: which are genuine deteriorations in fundamentals, and which are just emotional noise?
Three reasons why I am optimistic about it.
1. The data flywheel moat is unique in the AI healthcare sector.
This is not just empty talk. I compared it to all the competitors:
Foundation Medicine (Roche) – Only has genomic data, no AI platform. Flatiron Health (Roche) – Only has hospital electronic medical record data, no in-house genomic sequencing laboratory. Guardant Health – Only has liquid biopsy data, with limited coverage.
Only Tempus possesses multimodal data encompassing genomics, electronic medical records, imaging, and pathology, and its AI model is continuously fed data pipelines from over 4,500 hospitals.
Once this data barrier is established, it is extremely difficult to replicate. Although Google, Microsoft, and Amazon have general AI capabilities, they do not have hospital data pipelines or their own laboratories—they make "medical AI tools," not "medical data platforms."
2. The EBITDA inflection point has arrived.
Q3 2025 adjusted EBITDA turned positive ($1.5 million), reaching $12.9 million in Q4. Full-year guidance for 2026 is $65 million.
This means that Tempus is transitioning from a "cash-burning growth" phase to a "self-sustaining" phase. It has $760 million in cash reserves and has been running for over four years.
The gross profit margin of data services is 72-80% (SaaS-like), which is the core driver for future profit margin growth.
3. ARK voted with real money.
Sister Wood isn't just "optimistic"—she's expressing her opinion with money.
TEM is ARKG's largest holding (9.49%), ARKK's third or fourth largest holding (5%), and ranks fourth among all ARK funds. The total holdings are approximately 7.18 million shares, worth approximately $310 million.
On February 25, 2026 (earnings day), ARK bought $11.8 million worth of TEM in a single day. They continued to buy for several days in March. They even reduced their holdings in Meta, Roku, and Nvidia, transferring funds to TEM.
Wood has repeatedly stated publicly, "Healthcare is the most underestimated application scenario for AI."
ARK's judgment may not be correct. But when a fund manager managing hundreds of billions of dollars lists you as their largest holding, that's a signal worth taking seriously.
IV. Three Reasons That Worry Me
1. Securities Class Action Lawsuit – Five Charges That Should Not Be Underestimated
In May 2025, short-selling firm Spruce Point released a report, causing TEM to plummet 19% that day. Subsequently, a securities class action lawsuit was filed.
The lawsuit's five core allegations:
The contracts are overvalued (many are related parties or non-binding "intentions").
SoftBank's joint venture "repatriated funds" (SB Tempus contributed $65.3 million in revenue in FY2025, compared to only $4.5 million the previous year—did the money just go around in circles and become "revenue"?).
Ambry Genetics' aggressive billing
AstraZeneca's $200 million partnership was actually facilitated through a "bridging" payment via Pathos AI, a company affiliated with the CEO.
In summary, the prospects for the core business have been embellished.
The company responded, stating, "We believe the complaint is unfounded and we intend to actively defend ourselves." The case is still under review.
My assessment: Such lawsuits are extremely common among high-growth companies listed on the US stock market, with approximately 70-80% ending in small settlements. However, the real danger lies in the SEC issuing a formal investigation letter, or the auditor changing/issuing a qualified opinion. Neither has occurred so far.
2. Dual-class shareholding control by the CEO
Lefkofsky controls 64.3% of the voting rights through Class B shares (each with 30 votes). External shareholders have virtually no say in the company's governance.
He is the co-founder of Groupon—a company whose market value has plummeted from $16 billion after its IPO to less than $1 billion today. Spruce Point directly questions his historical pattern of "pushing disruptive technology companies, cashing out early, and leaving public shareholders with losses."
There were 118 insider sales and zero insider purchases in the past six months. The CEO himself reduced his holdings by over $34 million.
This is what makes me most uneasy.
3. The slowdown in growth is real.
The 83% growth rate in FY2025 was largely driven by Ambry's acquisitions. FY2026 guidance is 25%, with organic growth potentially at 20-25%.
For a company with a P/S ratio of 6, a 25% growth rate isn't bad—peers like Guardant's 21% growth corresponds to an 8-9x P/S ratio, and Veracyte's 20% corresponds to 7-8x. However, the market has woken up from its expectations of a "high-growth AI wonder stock," and a repricing is reasonable.
V. Valuation: Is $43 cheap?
I used 10 methods for cross-validation:
Method Implied Target Price
Comparison with peers (P/S): $71
EV/Revenue growth premium $60
Abbott/EXAS acquisition reference $66
DCF $42
Segment valuation: $53
PEG adjustment $63
Discounted forward P/E: $37
ARK implies a long-term $150+
With a median of approximately $58, the current price of $43 represents about 35% upside potential.
14 analysts: 8 buy/6 hold/0 sell. Consensus target price: $76-79 (implying +80%). Mizuho gives $100, JP Morgan gives $60, and Piper gives $55.
A P/S ratio of 6 is at the lower end of the industry average. Natera's is 12x, and Guardant's is 8-9x. If Tempus proves that the 25% growth rate is sustainable, a P/S ratio reversion to the mean of 7-8x would imply an upward trend of 30-65%.
But is $43 already a "half-price"? Not necessarily. DCF gives $42 (basically unchanged), and forward P/E discount gives $37. If the litigation escalates or the growth rate continues to deteriorate, the $35-$36 range may be the real bottom of the "emotional overreaction".
VI. Risk List
Escalating securities litigation → If the SEC formally intervenes, it could trigger a larger sell-off.
Related-party transactions continue to expand → If the proportion of AI revenue from SB Tempus/Pathos continues to rise, market confidence will decline.
Continuous selling by insiders → If there is unplanned accelerated selling, it is a serious signal.
GIPA Privacy Lawsuit → If the ruling restricts data licensing models, it will damage core business.
Growth rate below 25% guidance → If Q1 2026 falls short of expectations, repricing may be necessary.
VII. Conclusion
Tempus AI is in a classic "high odds but high uncertainty" window.
Reasons for optimism: The world's largest multimodal clinical data platform, EBITDA inflection point, 19/20 top Pharma clients, FDA-approved product portfolio, ARK's largest holding with counter-trend increases, and P/S ratio at the low end of its peers.
Reasons for a bearish outlook: securities class action lawsuit, allegations of related-party transactions, CEO's 64.3% voting control, 118 sell orders with zero buy orders, growth rate dropping from 83% to 25%, and short selling ratio of 18%.
This is not a ticket you can "buy and sleep on." It requires continuous monitoring of litigation progress, insider behavior, and quarterly performance.
But if you believe that AI-driven precision medicine is one of the biggest healthcare investment themes of the next 10 years—and I think it is—then Tempus is worthwhile to establish a small observation position in the $40-45 range.
Key catalysts: Q1 2026 earnings report (validating 25% growth), FDA approval of xF liquid biopsy, new collaborations with major pharmaceutical companies, and developments in securities litigation.
Good data, complicated people, time will tell.
#危机投资模型 #Tempus AI #美股 #A股 #交易心得 #股票交易 #投资理财 #趋势交易 #止损 #资金管理 #交易心理 #炒股必看 ⚠️: This article is for personal investment research sharing only and does not constitute any investment advice!
P.S.: Creating content is hard work, so please like and follow! I will continue to provide you with high-quality investment research analysis!