Data as of November 6, 2026. $IAM closed at $11.64 on November 5: basic market value $551.7 million on 47.40 million shares, treasury-method diluted $580.8 million on 49.90 million shares, enterprise value $142 million basic and $171 million diluted on $409.4 million of pro forma June 30 cash. The stock traded near $10 intraday on November 6. That is not a close, and we use it only as an illustration. IPO price $16.00, first trade October 15, peak close $23.81. All trial data carry an August 31, 2026 cutoff.
Disclosure: I am long $IAM at an average of $9.95, bought before publishing. This is research, not a recommendation. This is the first part of our Iambic Therapeutics series.
Iambic went public on a clean story: an AI drug discovery platform with Nvidia on the cap table, AbbVie and Takeda on the partner list, and a lead HER2 drug with a 36% response rate and a 41% brain response rate after almost everything else has failed.
Each of those phrases is true. Each one also has a second number sitting behind it, usually in the same filing, and the coverage we read mostly did not use the second numbers.
The 36% includes unconfirmed responses. The confirmed figure is 23%. The 41% is 18% confirmed. The IPO priced below the last two private rounds, the last private round missed its size by a third, and a director's fund bought $11.5 million of the IPO in a purchase the prospectus holder table does not show. Reported cash burn of $24.9 million in the first half of 2026 was $47.7 million before partner prepayments. Partners have paid about 2.8% of the disclosed headline deal value. And by the company's own attribution figure, the lead drug came from machine learning plus high-throughput chemistry, not from the two branded models in the equity story.
Here is the other side, and it is just as well documented. The drug produces real tumor shrinkage in brains that have already seen T-DXd and tucatinib, and only one of the 22 brain-evaluable patients had brain progression as best response. Four systemic responses and one brain response were still awaiting a confirmatory scan at the cutoff. Management hit nearly every dated milestone we could check. And the overvaluation argument is largely spent: the stock has fallen 27% from the IPO price to the November 5 close, and enterprise value has fallen from $349 million at $16 to $142 million.
This post does not reach for a verdict. It tries to put the right numbers on the table, label which ones are ours, and show what has to happen next. Where we infer, we say so. Every cross-trial comparison here is not head to head. Nothing in the record we reviewed suggests misconduct. Discrepancies are presented as discrepancies, with benign explanations where they exist.
Let's go.
The twelve findings, in one table
I. What Iambic is
Iambic Therapeutics was incorporated in Delaware in October 2019 as Entos, Inc. and renamed in May 2023 [1]. The founders are Tom Miller, a former Caltech chemistry professor, and Fred Manby, a former University of Bristol professor [1]. It is a San Diego clinical-stage oncology company with 171 full-time employees at September 18, 2026, up from 147 at June 30 [1][2].
Certificate transparency logs for the old entos.ai domain show names such as envision, foundry and sierra-docs [3]. We infer that the earlier business looked like a molecular simulation software company. The 2023 rename came with a therapeutics identity, and the Oct 2023 Series B funded that pivot.
The business has three parts: one clinical drug, two programs about to enter the clinic, and a discovery platform rented to pharma.
IAM1363 is an oral, covalent, brain-penetrant small molecule that inhibits HER2, including mutant forms, while trying to spare EGFR, whose inhibition drives the rash and diarrhea of older HER2 pills. The company describes it as binding the inactive DFG-out shape of HER2, a Type II mechanism, and says it believes it is the only known HER2 TKI to do so [1]. That is a company belief. We found no published co-crystal paper or PDB entry for IAM1363 itself. It is in a Phase 1/1b basket trial, with 122 people dosed at the cutoff [4]. The go-forward dose is 960 mg once daily. The company anticipates a registrational trial as early as 2027, subject to regulatory feedback [1].
Sources: [1][5][6]. The earlier public KIF18A compound was called IAM-K1, with a 10 nM OVCAR-3 IC50 [7]. IAM217 shows 24 nM [1]. We infer that some potency was traded for brain penetration, and IAM-K1 data should not be read across.
The stated development paths matter more than the pipeline chart, because they tell you which competitors count. Per the prospectus [1]:
Fastest path: IAM1363 alone in HER2-positive breast cancer with active brain metastases after T-DXd and tucatinib.
Primary breast path: IAM1363 plus trastuzumab or zanidatamab plus capecitabine in TKI-naive second- and third-line disease, with a potential registrational comparison against the tucatinib triplet.
Stomach and esophageal: with T-DXd against T-DXd in second line, or alone in third line.
Lung: with chemo and pembrolizumab in first-line HER2-amplified NSCLC, and alone in second-line-plus amplified disease.
Notice the softer language on HER2-mutant NSCLC, where two approved drugs have run ahead. The filing floats a first-line combination with an ADC and states a belief in brain penetration and selectivity, and names no registrational comparison there. We read this as a breast and CNS story first.
II. The confirmed number
Oncology response rates come in two flavors. A partial response, or PR, is at least 30% tumor shrinkage on one scan. A confirmed PR needs a second scan, usually weeks later, to show it held. FDA labels and pivotal trials report confirmed numbers. The registered primary endpoints of this trial are confirmed ORR and confirmed CNS ORR, with best overall response as a secondary [6][1].
Here is the same August 31 dataset in three documents.
Sources: [8][1][4].
The release does mention selected confirmed cases, including three confirmed intracranial PRs in HER2-positive breast cancer after both T-DXd and tucatinib [8]. What it leaves out is the confirmed rate itself. The rate was public: it sat in the S-1 from September 21 [9], so nobody buying the IPO was denied it. The coverage we read mostly ran with 36 and 41 [10].
The arithmetic of what is pending. We did this math ourselves.
Systemic: 19 PRs, less 12 confirmed, less 4 pending, leaves 3 that are neither. We infer those failed to confirm or lack a confirmatory scan. If all 4 pending confirm, the rate is 16/53, or 30%.
Brain: 9 PRs, less 4 confirmed, less 1 pending, leaves 4. If the pending one confirms, the rate is 5/22, or 23%.
SD-approaching-PR cases (2 systemic, 3 brain) could add a little more over time.
Exact 95% confidence intervals, also ours: systemic confirmed 22.6% (12.3 to 36.2), systemic best response 35.8% (23.1 to 50.2), brain confirmed 18.2% (5.2 to 40.3), brain best response 40.9% (20.7 to 63.6). These are small samples. An honest reader should hold 23% loosely in both directions.
The finding we have not seen anywhere else: the prior extract. The confidential drafts, now public on EDGAR, carry an earlier cut. On the May 18, 2026 extract, the July DRS reported 34% best response (14/41) and a 27% confirmed rate (11/41) with one PR pending and two near-PRs, plus 39% brain best response (7/18) and 17% brain confirmed (3/18) with a 100% intracranial disease control rate [11][2]. Lay it next to the older and newer data.
Sources: [12][11][2][1][4].
In 15 weeks the denominator grew by 12, best responses grew by 5, and confirmed responses grew by 1. The headline went up. The confirmed rate went down.
There are good reasons not to over-read this. These were confidential drafts, so this is no public retreat by the company. New patients have not had time for a second scan, which is exactly why pending PRs rose from one to four. If all four confirm, the rate is 30%, which is above the May figure. Population and cut definitions may differ slightly between extracts. The right conclusion is narrow: the 34-to-36 climb and the 27-to-23 slide describe the same patients at different stages of maturity, and the next confirmed number is the most important figure in the company. A mature rate above 30% changes the conversation. A mature rate in the low 20s also changes it, the other way.
Three more things the headline does not tell you.
First, the denominator. 53 is about half of the 103 people at 960 mg or above. Enrollment ran on local pathology. People whose tumors central review found lacked a targetable HER2 alteration were dropped from efficacy but kept in safety [1]. That is a defensible way to measure a targeted drug. It also means the rate is not an intent-to-treat number, and the exact attrition is not published.
Second, the dose. The efficacy pool is 960 mg or greater, including 1200 mg, 1560 mg and 600 mg twice daily. The cleanest safety figures are 960 mg alone. We have found no disclosed response rate at 960 mg alone, the dose going forward. The company says 960 mg and 1200 mg performed equivalently, and the poster reports no PK advantage at 1200 mg [4]. This is the single most important missing number. Each figure is labeled correctly, so we would call this framing, not misstatement.
Third, the brain criteria. The poster reports modified RANO-BM with target lesions of at least 5 mm. The prospectus text says RANO-BM [4][1]. The standard criteria, from our background knowledge, use a 10 mm threshold for measurable lesions. Smaller lesions are harder to measure, and that can move a response rate either way. It is a reasonable choice in Phase 1, and worth watching in a registrational design.
The number that decides the fastest path. 19 of the 22 brain-evaluable patients had HER2-positive breast cancer. 21 had prior T-DXd and 19 prior tucatinib [1]. The poster reports 3 confirmed brain PRs in HER2-positive breast cancer, all after both drugs and all still on treatment, 2 beyond six months, plus one confirmed brain PR in HER2-mutant NSCLC [4]. So we infer the confirmed brain rate in the population Iambic calls its fastest route is about 3 of about 19, or roughly 16% (95% CI 3.4 to 39.6), possibly 4 of 19 if the pending case is breast.
Is that good enough for a single-arm accelerated approval? There is no approved therapy in that setting, so the bar is lower than in lung. It is also not obviously met at this cutoff. Duration will decide it, and no median duration of response or PFS has been published. The durability evidence so far is individual: a HER2-mutant NSCLC confirmed PR ongoing beyond 12 months, 3 patients on treatment more than a year, 12 more at six months or longer, and 40 of 122 (33%) still on treatment overall [4][1]. PubMed returns zero records for IAM1363, so none of this has been peer reviewed as a full paper [13].
III. Thirteen versions of the registry
The trial is NCT06253871, plus EU record 2024-519188-17-00. We diffed every ClinicalTrials.gov version for status, enrollment, dates, arms, outcomes, eligibility and sites [14].
Net, primary completion is four months earlier than first registered. The December 2024 change looks like cleanup. April 7, 2026 is the one that matters. That single version:
Added Part 4: capecitabine plus trastuzumab, capecitabine plus zanidatamab, T-DXd, and pembrolizumab with or without carboplatin and pemetrexed.
Took enrollment from 243 to 383, up 58%, and moved primary completion from March to December 2027.
Added a new exclusion for prior non-infectious ILD, with resolved Grade 1 allowed.
Removed the part limits on confirmed ORR and CNS confirmed ORR, so both became all-part primaries, and moved dose modifications, labs and ECG into primary outcomes.
Added central HER2 testing for selected cohorts and 14-day or 21-day cycles [14].
We read that as scope expansion, not an enrollment failure. The ILD exclusion needs more care. It arrived in the amendment that added T-DXd, a drug with a known ILD risk, so the combination alone could explain it. Separately, the safety extract dated one week later, April 14, 2026, listed a Grade 3 ILD event at 960 mg [11]. No document states why the exclusion was added. It could be either, or both.
Other details from the record:
Dose drift. Version 2 planned 120 to 480 mg. The trial tested up to 1560 mg, and 960 mg was chosen. The go-forward dose is twice the original planned maximum [14][12]. We infer this matters for pill burden in combinations.
Enrollment reconstruction, our estimate. With 122 dosed, Part 1 at 30, Part 2 at 45 planned and the poster's dose counts, we estimate Part 3 at about 34 of 104 and Part 4 at about 13 of about 115 at the cutoff. If a combination-based pivotal starts in 2027, it will lean on small early cohorts.
Unexplained capacity. Public cohorts add to about 292-294 against 383 registered. About 90 slots are not explained by anything public.
Stale language. The current record still says 46 months for completion and refers to a Simon two-stage design [6].
Site footprint. 53 facilities: US 29, Spain 6, Italy 4, Korea 4, France 3, Ireland 3, UK 3, Netherlands 1 [6]. Authors moved from Phase 1 networks in 2025 to a list led by MD Anderson and including Dana-Farber and Vall d'Hebron in 2026 [12][4]. That is a real upgrade in who is running the drug.
Registry mismatches. ClinicalTrials.gov shows all sites recruiting, including START Madrid, which START's own page lists as on hold [15]. CTIS gives Italy's start of recruitment as October 20, 2026, while the US record listed Italian sites as recruiting from January to June. CTIS names Singapore as a third country with no Singapore site on the US record [16]. We read these as registry lag and site-level management. CTIS shows no trial-level halt, urgent safety measure or serious breach.
Cohort sizes. Cohort 4D is planned at 26 on the poster and 24 in the prospectus [4][1]. Small, but the 383 is built from these.
IV. Poster versus prospectus on safety
IAM1363's tolerability at 960 mg looks acceptable for a late-line oncology pill. Grade 3 treatment-related events hit 4 of 68 (5.9%) at 960 mg against 6 of 29 (20.7%) at 1200 mg. Discontinuation for treatment-related events was 0 at 960 mg against 2 at 1200 mg [4]. That gap supports 960 mg as a real dose selection rather than a formality.
Sources: [4][1]. Note the mix: efficacy is reported at 960 mg or above, the best safety line at 960 mg alone and treatment-related only. Each is labeled.
We compared the documents and found three differences. They may all have benign explanations. They are worth a direct question to the company.
On liver, the April draft reported no treatment-related AST or ALT elevation at 960 mg [11], and the poster shows a Grade 3 AST event in 1 patient at 1200 mg [4]. The press release does not mention pneumonitis [8]. One case in 68 cannot define a rate. Two across the 97 patients at 960 mg and 1200 mg show that a clean ILD claim is not available.
What remains undisclosed: G1-2 pneumonitis or ILD rate, ALT and AST rates under 10%, rash rate under 10%, anti-diarrheal prophylaxis, QT and LVEF results, half-life and human brain exposure [4][1]. Biopsy data for tumor enrichment come from only 2 participants [1].
V. 1,000-fold, 5,000-fold, and the prospectus
Selectivity over EGFR is the core design claim. The figure has moved.
We searched the raw 424B4 text. ">1,000-fold" appears twice, and 5,000-fold does not appear. The posters and prospectus were live in the same month. A likely benign explanation is a different assay or calculation. No document we read explains it. We would not repeat 5,000-fold without a method.
Clinical evidence for EGFR sparing is indirect and fairly good: rash does not reach the 10% table at 960 mg. Selectivity does not make the GI profile clean, however. Diarrhea, nausea and vomiting are all common. Whether they are on-target, off-target or from pill load is not addressed.
The poster's claim of 10-fold greater brain penetrance than tucatinib is also not in the prospectus text, which gives no such ratio or human value [4][1].
The registry and FDA language is similarly modest. The prospectus mentions Breakthrough and Fast Track only in generic terms, names no designation for IAM1363, and describes FDA interaction as ongoing engagement [1]. EMA's orphan list has no hit [21]. No meeting type or written FDA feedback is disclosed. FDA's own orphan database blocked our query, so that one is unverified.
VI. The IPO was a down round
$16.00 was the midpoint of the $15-17 range [1]. It was also below what private investors paid in the prior 12 months. The return table is ours, using filed prices [1].
So $16 was 1.9% below B-3 and 6.7% below B-4. About 19.4% of post-IPO shares were bought above the IPO price within the prior year. At $11.64, every series other than Seed and A-1 is under water. Existing holders average $12.45 per share, so the stock now sits below pre-IPO holders' average cost [1].
The $13.60 notes. In August and September 2026, weeks before the public filing, the company sold $66.5 million of 8% subordinated notes that converted at the lower of 85% of the IPO price and a price implied by a $900 million valuation cap. Form 4 footnotes confirm 85% of $16, or $13.60 [1][22][23]. The prospectus describes the conversion as based on the $16 IPO price, which is how the share count is set. The 4,949,237 shares times $13.60 equals $67.3 million, which ties to principal plus about $0.8 million of interest. Our arithmetic puts the transfer at about $11.9 million at the IPO price.
New names came in here: accounts advised by KKR subsidiaries, Insight Partners, Perceptive, Millennium and Laurion Capital [1]. Insiders took about $17.7 million of the notes: Catalio $10.0M, Ascenta $5.5M, Abingworth $2.1M and the chairman's family trust $0.1M. CFO Michael Secora spent 10 years at Laurion until 2020 [1], a normal small-world overlap we note for completeness. A pre-IPO crossover note is standard. The 15% discount tells you what it took to clear in the last weeks before going public.
The B-4 round missed its size. The purchase agreement allowed for up to 2,916,580 shares, or about $50 million, and the April 2026 Form D offered $49,999,996 [1][24]. The company sold 1,911,506 shares for $32.8 million [1]. Common reserved for preferred conversion is 29,530,003 against 28,524,929 outstanding. That gap of 1,005,074 is exactly the unsold B-4 count. We infer about $17 million of the target was never placed.
Now see who bought. Catalio Access Fund VIII took $25.0 million, Nvidia $5.0 million and the Rastetter Family Trust $0.257 million, which is $30.3 million of $32.8 million [1]. Catalio's co-founder Jacob Vogelstein sits on the board, and William Rastetter chairs it. We infer outside money in the last equity round was nearly nil. Three months later the company was selling discounted notes.
Read charitably, this is committed insiders backing the company at a price above where the IPO cleared, and they bought before an IPO that was not yet assured. Read skeptically, the last priced round was not set by new money. Both are fair.
Ascenta bought in the IPO, and the table does not show it. A Form 4 filed October 19 under director Evan Rachlin reports Ascenta Capital Fund I buying 718,750 shares at $16.00 on October 16, transaction code P, or $11.5 million, 7.7% of the deal [23]. The 424B4 holder table, as of June 30 and excluding IPO purchases, shows Ascenta at 4.1% post-offering [1]. Our count after the purchase and note conversion, including the Project Turing SPV line, is 2,646,002 shares, or 5.6%. So a holder the table shows under 5% looks over it, and a 13G or 13D should eventually say so. No other Form 4 shows a code P purchase. No Form 4 shows a sale [25].
"Nvidia-backed" is a 3.9% holding. Nvidia bought $20 million of Series B and $5 million of B-4, or 1,827,692 post-split shares [1]. The S-1 listed Nvidia, Q Healthcare and Coatue as greater-than-5% holders. The S-1/A table removed them [9][26]. We infer the note shares enlarged the denominator: Nvidia is 4.8% pre-IPO and 3.9% post, and nothing suggests anyone sold. Nvidia also co-owns one patent family and has a technology collaboration on its hardware. We found no collaboration payment from Nvidia to Iambic.
The auto-convert price. Preferred mandatorily converted only at an IPO price of at least $39.07, or on a holder vote [1]. $16 is well below that, so we infer holders voted to convert. The notes at $13.60 and the IPO at $16 are below B-2, B-3 and B-4 conversion prices, and the count shows no anti-dilution adjustment, consistent with the waiver mechanism in the charter terms. We did not find the consent text.
Common stock. Board-set common fair value was $8.94 in April through June 2026, 52% of the B-4 price sold in the same months, then $12.47 on August 7 [1]. A rising common-to-preferred ratio is normal into an IPO. The grants at those levels are small enough that we estimate any cheap-stock charge at under $6 million over vesting.
The tape. First close $18.91, then $23.81 on a $25.00 high on day two, then a 31% one-day fall to $16.34 on October 19, below IPO from October 21, $11.64 on November 5 with a $10.55 low [27]. On November 6 at our cutoff, it was about $10 with a $9.60 low. That is not a close [28].
We found no filing or news item explaining October 19, November 5 or November 6. XBI fell 6.7% on November 5 by our Yahoo data, so that day was at least partly sector [29]. Since the listing date, XBI is about flat at the November 5 close while IAM is down 27% from offer. Volume totals 1.48 times the IPO share count in 16 sessions [27]. With nearly all pre-IPO holders locked and no Form 4 sales, we infer the selling came from IPO-era buyers, which public data cannot prove.
VII. Cash, burn and what the screens get wrong
Stock screens show $207.9 million of cash, $18.25 million of revenue, -$65.7 million of operating cash flow and 47.4 million shares. All four tie to filings. All four mislead if you take them as current [1].
Reported versus underlying burn. First-half 2026 operating cash flow was -$24.9 million. That figure includes a $33.2 million inflow from deferred revenue and a $10.3 million outflow into receivables [1]. Those are mostly partners paying ahead, and we infer Takeda's $27 million upfront is the biggest piece. Strip both and you get our underlying series.
Reported burn appears to fall 39% from H2 2025 to H1 2026. Underlying burn rose 25%. Net loss rose too, from $33.2 million to $50.1 million year on year [1]. No quarterly data exists in any filing, so the first real quarter arrives with the first 10-Q.
Within R&D, H1 2026 ran $17.4 million on IAM1363 external costs, $19.1 million on discovery and platform and $17.6 million on people. The segment note gives a line most writeups miss: compute costs of $4.36 million, up from $1.71 million a year earlier [1]. IAM217 and IAM-C1 are not split out.
The runway claim. The prospectus says existing cash plus net proceeds fund operations and capex into or through the fourth quarter of 2028 [1]. That date was blank until the S-1/A [9][26]. On our estimate of about $418 million today, reaching December 2028 implies about $200 million a year, roughly double the $95 million annualized underlying H1 burn. Either the plan ramps sharply, which a Phase 3 would explain, or the guide has cushion. The company also says partnership proceeds should extend runway, and does not say which ones are in the plan.
Use of proceeds is more specific than the summary suggests [1]:
About $75 million for IAM1363: finish Phase 1, start one or more Phase 2 trials for possible accelerated approval, and start a Phase 3. Start, not finish.
About $15 million each for IAM217 and IAM-C1, through the end of Phase 1 dose escalation only.
About $20 million for the platform: model training, lab automation, agent software. Against $8.7 million of annualized compute, that is about 2.3 years if compute were the only draw.
The remainder of about $10 million is working capital, by our subtraction.
We infer a further raise before pivotal data is the base case. The company says it will need substantial additional capital [1]. In our view a randomized PFS readout by Q4 2028 looks unlikely from a 2027 start.
The rent holiday. The new San Diego headquarters at 3545 Cray Court is a 7.7x jump in space. Iambic holds about 12,545 sq ft on Oberlin Drive through June 2031. It signed about 45,000 sq ft in January 2026, amended to about 58,400 sq ft in March, then superseded that with about 96,000 sq ft on July 17, 2026, one week before the first confidential filing [1].
The original premises carry $63.00 per rentable square foot with 3% annual increases, over a 90-month term, and Section 3 of the lease waives base rent for the first 32 months [30]. The filing confirms 32 months on the original space ($15.7 million total commitments) and 10 months on the expansion ($25.0 million), about $40.9 million in all [1]. Our arithmetic: 44,912 sq ft at $63 is about $2.83 million a year, so the holiday is worth about $7.5 million. Target commencement of November 2026 means base rent on the original space starts about mid-2029, after the stated runway. The lease flatters near-term burn. Whoever funds 2029 pays it. On 171 staff the footprint is about 561 sq ft per head, against about 73 today, so it is a bet on substantial growth, more automation, or both.
The Alexandria overlap. The landlord entity is ARE-SD Region No. 38, LLC, and the project is One Alexandria Square [30]. Alexandria Venture Investments is named in Iambic's November 2025 financing release [31], two months before the lease. We infer the landlord and investor are affiliates. The word Alexandria does not appear in the 424B4 text, including the related-party section, and we have no evidence that any disclosure rule required it. Vendor-investor overlaps are common in life-science real estate, and we draw no conclusion beyond noting the link. Wilson Sonsini, deal counsel, also appears on that financing list [31].
The Irish money. Ireland Strategic Investment Fund lists a $20 million commitment to Iambic under 2026, and its April 2026 newsletter ties it to an engineering and technology presence in Ireland [32][33]. Iambic Therapeutics IRE Limited was registered May 5, 2026, and is one of exactly two subsidiaries [34][35]. Iambic leases a month-to-month Dublin office [1]. "Ireland Strategic" appears nowhere in the 424B4, and we could not map the $20 million to a specific round. We infer it carries some local-presence expectation; terms are not public. It is the only government-linked capital we found. NIH RePORTER and USAspending show no award to Iambic or Entos, Inc., and lobbying records and Open Payments show nothing [36][37][38][39]. Do not confuse this company with Iambic Inc., a footwear NSF grantee, or Entos Pharmaceuticals, an unrelated Edmonton company.
A clean balance sheet, at least. No borrowings. Service contracts are cancellable on 14 to 120 days' notice, with no material non-cancellable obligations outside leases. The financial statements have no substantial-doubt language [1].
VIII. What the platform has and has not proven
Start with the most useful figure in the filing. Figure 2 of the prospectus lists how platform technologies were used on each program [1].
The branded models get credit on the two programs with no human data. The drug with human data is credited to machine learning plus automated chemistry. To be fair, the Business section credits the molecular superintelligence platform broadly for IAM1363's properties [1], and active learning over an automated lab is genuinely AI-driven chemistry. But timing supports the narrower reading. The earliest HER2 patent priority date is August 22, 2022, the IND was announced December 2023, and Enchant v1 training completed in 2024 [40][17][1]. We infer Enchant could not have designed IAM1363.
HER2 is also the easiest place to post a fast cycle time, with decades of chemotypes, validated assays and an obvious clinical path. So the "about two years, versus about five and a half" claim is real, and weak as evidence about generality. The baseline comes from an unnamed third-party analysis plus internal work [1]. Conventional competitors reached approval while IAM1363 remains in Phase 1. IAM217 is the better test, and its IND went in on September 30.
NeuralPLexer has a real paper and an unreviewed sequel. NeuralPLexer 1 is peer reviewed in Nature Machine Intelligence, with 178 citations by OpenAlex, open code and data, and independent reuse [41][42]. NeuralPLexer3 reports 78.4% against AlphaFold 3's 73.1% on the PoseBusters benchmark, in an arXiv paper with no code or weights [43][44]. We found no third-party benchmark of NeuralPLexer3. The leading independent comparison, PoseBench, covers AlphaFold 3, Chai-1, Boltz-1 and the older NeuralPLexer, not version 3, and its own rankings shifted between versions [45]. A 5-point self-reported gap sits inside that noise. We infer it does not establish durable superiority, and that open models are commoditizing co-folding at a pace Iambic cannot match on research budget.
Enchant has no paper. We found no preprint or peer-reviewed publication, only company posts [46][47]. The v1 headline is a Spearman of 0.74 against a prior 0.58 on the public Obach human half-life benchmark. The post discloses no split method, no dataset size and no leakage control. Version 2 reports wins against a fine-tuned TxGemma, including rat clearance AUC of 0.70 against 0.35 for a hepatocyte assay, on undisclosed internal data. The filing gives scale: 1 billion parameters, about 7 billion, then 41 billion for v3 launched in 2026 [1]. It gives no v3 benchmark numbers, and its LLM comparator in Figure 10 is unnamed. The September 21, 2026 launch of v3 shared a date with the public S-1 [48].
What the record shows is PK and ADME regression, some of it impressive. The record does not show prediction of clinical efficacy or trial success. The only paper we found on that subject is a July 2026 Research Square preprint from Iambic authors [49]. The July 2026 DRS called the platform thoroughly validated, and that sentence was gone from the August DRS/A. The DRS also said the platform can systematically improve cost, time and probability of success; the DRS/A and everything after say it has the potential to [11][2][1]. Both changes happened at the confidential stage, before any public filing, so neither is a public retreat. Counsel or SEC comment is the likely cause, and no letters are public yet. The company itself calls its approach novel and unproven [1].
Moat by asset.
We infer the platform moat is a data-and-execution moat, not a patent moat, and it sits on the layer with the least public evidence. Verifiability has declined as commercial importance has risen. Both decisions are rational. Investors bear the consequence.
The people picture is mixed but not alarming. Fred Manby and Tom Miller remain. The authors behind the models are mostly not named in the filing, which is normal for non-officers. The about page lists Matthew Welborn as SVP, ML [50]. We could not establish the status of first author Zhuoran Qiao, which we would ask about. Compute runs on Lambda and AWS, with Lambda's cluster supporting Enchant training per the filing [1][51]. The filing describes optimizing models for Nvidia hardware and does not name Nvidia as a cloud provider. We found no DGX disclosure and no material compute commitment.
IX. Headline value versus cash
Partner names are the second pillar. Here is what they have actually paid [1][52][53][54].
Disclosed headline on the three deals with numbers is about $2.66 billion: Takeda $1.727B, AbbVie $710M, Lundbeck $222.5M. Cash from all partners through September 2026 was $74.7 million, or about 2.8% [1]. If you add AbbVie's $40 million, which we infer is not in the $74.7 million, you get $114.7 million, or about 4.3%. One passage dates the $74.7 million as of September 18, the day AbbVie signed, and another as through September 30. Cash implied through June 30 is about $65.3 million by our math, leaving about $9.4 million for Q3. The first 10-Q should settle it.
Biobucks gaps are normal, so here is the sharper version:
Eight partners is really five discovery deals. "About eight partnerships in about eight quarters" [55] includes a compute vendor, a chip maker that is also a shareholder and a clinical supply agreement.
Bayer is small. Remaining transaction price of $4.0 million including estimated reimbursement caps the upfront around that level, by our inference.
Remaining price nearly equals deferred revenue. $53.4 million against $53.6 million. Lundbeck, the largest revenue contributor to date, has $3.8 million left and a research term ending by about September 2027.
No Takeda, Lundbeck or Bayer milestone had been hit at June 30. RevMed is the only one.
Options can lapse. Lundbeck's and AbbVie's later economics depend on exercise. Takeda is a direct exclusive license.
Redaction. We count 784 redaction markers in the AbbVie exhibit, 475 in Takeda's and 301 in Lundbeck's [54][52][53]. Bayer, RevMed and Jazz are not filed at all, which we infer means they were judged immaterial under Item 601.
Takeda data does not all compound. The Takeda agreement requires Iambic to delete and expunge program-dedicated models and Takeda datasets within 30 days of a research program's end, and certify on request (Section 2.4.2) [52]. That sits awkwardly with a story in which partner work keeps teaching the platform. Terms vary by deal: Lundbeck lets Iambic use data for training [53].
AbbVie sequencing. The deal was signed September 18, three days before the public S-1, and the company's first public filing led with it. That is good execution. It also means the largest upfront arrived in time for the roadshow.
The counterparties barely mention it. RevMed gives it a paragraph with no dollars [56]. AbbVie gives it one line in its Q3 release, in a quarter with $216 million of acquired IPR&D [57]. Takeda's 20-F hits on the name, though we could not read the passage [58]. We infer the relationship is highly asymmetric: central to Iambic, immaterial to them.
The deals are real, with real cash and substantial rights. Three pharma signings in eight months is genuine demand. None of them touches IAM1363, and none yet shows a partnered candidate reaching the clinic. Lilly is not a partner in any document we read; the name has zero hits in the prospectus. Validation of the discovery engine is partial until a partner exercises an option or doses a patient.
X. Single source, China and the word "redundant"
Two passages in the same prospectus [1]:
Business section: the company manages a diversified CDMO network to establish "secure, redundant, and highly compliant" supply chains across the portfolio.
Risk factors: the company relies on a single contract manufacturer for IAM1363 drug substance and a limited number for drug product, and has no qualified second-source supplier for drug substance. A significant portion of third-party manufacturing is concentrated with manufacturers in, or affiliated with entities based in, China.
These can be reconciled. "Designed to establish" is aspirational language, and the risk factor governs the specifics. Even so, a reader of the Business section alone would get a different impression. The risk factor also says the company does not have redundant suppliers for all materials and programs, so the filing corrects itself.
WuXi AppTec is the only named vendor, for certain R&D services, and the filing flags it as high risk for BIOSECURE designation because of its 1260H inclusion [1]. We do not assert WuXi makes IAM1363, because no filing says so. CDMO names are not disclosed.
The law: BIOSECURE was signed as NDAA Section 851 in December 2025. It restricts federal contracts, grants and loans where a listed company's services are used. OMB's list is due by mid-December 2026, and grandfathering runs five years [1][59]. Iambic has no federal funding that we can find, so we infer direct exposure is low. The indirect exposure is partners pressing for non-China supply, plus migration cost and timing before a registrational program.
One wrinkle runs in the company's favor. Secondary reporting and law-firm analysis say a D.C. federal judge preliminarily enjoined WuXi's 1260H designation on August 7, 2026 [60]. The October 14 prospectus does not mention an injunction. That makes the risk description conservative, not understated. We read this from search text and did not confirm appeal status.
XI. Organizational exhaust: hiring, offices, scorecard
Hiring. Iambic's public job board and archived snapshots give ten dated views [61][62]. The classification is our keyword scheme.
Across 92 unique postings, a lower bound, we saw the shape of a pre-registrational build arrive mostly in Q3 2026: head of medical writing, regulatory CMC, clinical pharmacology, drug safety with Phase 1 to 3 experience, a lead data manager with pivotal experience, quality systems and drug product. That fits "as early as 2027." We infer it also leaves a tight runway if the start is early rather than late in that year.
Zero commercial roles. No commercial operations, market access, sales or medical science liaison title appears in any of the 92. The closest is a post-IPO Corporate Counsel, Commercial posting, which is a contracts lawyer. That fits a company with no launch in sight, as you would expect in Phase 1. Postings did add immunology, clinical-prediction ML and agentic AI in Q3 2026, which fits management's comments on moving beyond oncology.
Headcount rose 24 in the 80 days to September 18. Twenty-two were in R&D [2][1]. California's current WARN file has no Iambic entry [63]. We have no attrition data, and page removals on the about page coincide with a redesign, so we would not read them as departures.
Footprint and records.
Bristol is small on paper. Iambic Therapeutics UK Ltd, formerly Entos AI Ltd and renamed March 2026, had an average of 5 staff in FY2025 and recognizes revenue under a transfer-pricing agreement with the parent; Manby is sole director [64][65]. A cost-plus unit relying on parent support is routine.
EPA. Two RCRA handler IDs, one each at Oberlin and at Cray Court, with no inspections, violations or enforcement [66][67]. The Cray Court ID being live ahead of commencement fits move preparation.
Auditor. EY was dismissed and Deloitte appointed on September 4, 2025, for FY2024 and FY2025. No disagreements, EY's report was unmodified, and one reportable event, a material weakness identified in 2023, remediated by December 31, 2024 [1][68]. A change of this kind ahead of an IPO is common, and the filing's own timeline fits IPO preparation. The filing does not describe the weakness's nature.
Courts. The company reports no material legal proceedings [1]. We could not query docket indexes.
Team. CMO Neil Josephson joined January 2024 after Zymeworks and Seagen. Iambic credits his work on zanidatamab, a HER2 bispecific and now a combination partner in Part 4, and we note his HER2 record is antibody, not TKI [1][69]. CSO Peter Olson came via Mirati and BMS. CFO Michael Secora ran finance at Recursion from 2020 to 2024. Chairman William Rastetter is a veteran of IDEC and Neurocrine. Samit Hirawat, BMS's CMO until November 2025, joined the board in August 2026 [1]. It is a credible public-company bench assembled in stages from 2024.
Incentives. Miller and Manby took 54% of IPO options and 44% of IPO RSUs, at a $16 strike that is 27% under water at $11.64 [70]. Their 634,788 PSUs vest on first dosing in a HER2 trial that would support approval, on a US or EU HER2 partnership, or on a change of control, by March 2033 [71]. We infer founders have some reason to partner or sell IAM1363, and that is not a bad thing for outside holders. The Phase 3 trigger is well aligned.
The website, over time [72][73][74][75][76][77][78].
We infer legal review softened the superlatives, which is normal before an IPO. The redesign likely explains the partner grid. We sampled captures and cannot tell whether all hidden elements displayed. "Superintelligence" is a marketing term the filing does not define, including in its financial statement notes.
Promise versus delivered.
Sources: [79][80][17][81][82][31][26][55]. On dated, checkable items, management has generally delivered. The claim language has tended to run ahead of the evidence, which is a separate question.
XII. Competition: every comparison here is cross-trial, not head to head
No randomized comparison of IAM1363 with anything exists. These trials differ in tumor type, line of therapy, prior exposure, assessor and follow-up. IAM1363's pool is a heavily pretreated basket with a median of 3 prior metastatic regimens, 61% prior T-DXd and 36% prior tucatinib [4]. Comparisons can flatter or punish it. We show them because a prescriber and an FDA reviewer will do this anyway.
HER2-mutant NSCLC and class safety. Not head to head.
Sources: [83][84][85][86][87][4][1]. Zongertinib figures use the July 2026 label. IAM1363's numbers are a mixed population, and no NSCLC-only rate is published.
What we take from it. Lung is largely closed in first and second line, with experts saying no evidence supports within-class TKI switching for true progression [88]. A rash rate under 10% is the cleanest clinical hint of EGFR sparing, against 32% and 66%. On a like-for-like all-cause basis, diarrhea does not separate IAM1363 from zongertinib, and nausea at 53% exceeds zongertinib's 25% [83]. Against sevabertinib, the tolerability gap looks real.
One competitive claim deserves a flag. The filing says neither zongertinib nor sevabertinib meaningfully crosses the blood-brain barrier, judged by disclosed Kp [1]. The Hernexeos label shows RANO-BM responses in 3 of 5 patients with measurable CNS disease [83], and a Sept 2026 OncLive panel cites an IC-ORR of 47% in active brain metastases in first line (n=30), while also calling intracranial control the central unsolved gap, at around 30% to 50% [89]. A Kp argument is not a clinical argument. Iambic's confirmed brain data is mostly breast, after two prior agents, so it cannot be set against those lung numbers either.
HER2-positive breast cancer. Not head to head. Those trials enrolled patients not refractory to both T-DXd and tucatinib, as far as we understand, so they overstate the gap.
Sources: [90][91][92][93][94][95][4].
Roche's BREnnA is the trial to read. NCT07413939 is a Phase 2/3 of RO7771950 against tucatinib, each with trastuzumab and capecitabine, in post-ADC, TKI-naive HER2-positive metastatic breast cancer with brain metastases allowed, about 650 patients, recruiting, primary completion May 1, 2029 [96][97]. Roche says the drug has higher CSF penetration. That is the same comparator and essentially the same setting as Iambic's stated primary breast path. Roche is randomized and enrolling. Iambic is hoping to start as early as 2027.
The prospectus's claim of no near-term competitors applies to the fastest path, active brain metastases after T-DXd and tucatinib. BREnnA does not test that exact setting, so the sentence stands [1]. The same filing lists RG-6596 from Roche and Zion among competing HER2 TKIs. We infer RO7771950 may be that drug, and could not confirm it. Dana-Farber names ZN-A-1041 alongside IAM-1363 as brain-penetrant HER2 pills [98].
Other trials worth tracking, per registry dates:
Sources: [99][100][101][96][102]. ELVN-002 from Enliven has been discontinued, so old lists overstate that one [103]. We found no oncologist commentary specifically on IAM1363, and no prescriber survey.
XIII. Commercial sizing: our estimates, with assumptions shown
Everything in this section is our estimate. These numbers are not guidance, consensus or precise. The company gives no peak-sales number. "A" marks an assumption of ours, and all patient counts are US and annual. We assume $18,000 net per month (A), against Tukysa's WAC of about $26,400 a month before gross-to-net [104].
Calibration. Tukysa did $402 million globally in the first nine months of 2026, $296 million in the US, for about $536 million annualized [105]. Enhertu did about $1.71 billion in Q3 alone, about 12 times Tukysa's $142 million [106]. HER2 small molecules are a far smaller franchise than HER2 ADCs. At about $165,000 net per patient (A), Tukysa's US run rate implies about 2,400 starts a year across all indications.
Inputs: [107][108][109][110][111][112][113]. The filing cites a higher brain metastasis incidence of 40-50% than the 23-34% we used from an 18,075-patient real-world study [1][113].
Read it this way. The bull case means roughly matching Tukysa, a drug with six years on the market and overall survival data. Base is a real but modest product. Bear is a niche late-line drug. The biggest and shakiest row is the TKI-naive breast slot: HER2CLIMB-05 and DESTINY-Breast09 push tucatinib and T-DXd earlier, so that pool should shrink while the post-both pool grows, and BREnnA is aimed right at it. Unrisked means no failure discount. One reported analyst estimate for Hyrnuo of only $23 million in 2028 and $119 million in 2032 is a warning on lung size, from a secondary source we did not verify [114].
We did not build value for IAM217 or IAM-C1 here. Both are preclinical until first dosed.
XIV. Cap table, float and the April unlock
One share, one vote, with a classified board and standard defenses [115][1]. Officers and directors as a group hold 21.7% post-offering by the table [1].
Sources: [1][9][23][116][117].
The float is smaller than it looks. The IPO put 9,375,000 shares in public hands, 19.8% of the company. Ascenta took 718,750. ARK and Duquesne indicated up to $60 million on non-binding terms, or up to 3.75 million shares, 40% of the base deal [1]. Add those at the ceiling and only about 4.9 million shares were left for everyone else. Actual allocations are not disclosed, so that is the low bound, not a fact.
ARKG's 1,665,353 shares equal 17.8% of the deal. The file does not show cost or source, so IPO allocation and open-market buying cannot be separated [116]. ARKK's file has no IAM line. ARKG plus ARK Venture is about 2.40 million shares, or about 5.1%, a threshold crossing worth watching. ARK Venture's marks tell a small story of their own: $16.30 at cost, $17.00 at July 31, 2026, about 6% above the IPO price set 11 weeks later [117][118].
The unlock. The lock-up is 180 days for directors, officers and substantially all holders. The schedule shows no staggered or earnings-based early release. About 38.0 million shares, 4.06 times the IPO float, become eligible beginning 181 days after the prospectus date, which we count as April 13, 2027, with 9,585,280 of those held by affiliates under Rule 144 limits [1]. Demand registration rights start about then, need a majority request covering at least 40%, and allow two demands [1]. The four banks can release early at their discretion, and a release for larger holders must be pro rata. We saw no sign of one.
History says this is usually not a cliff. Field and Hanka found a three-day abnormal return of -1.5% across 1,948 lock-ups, with a larger drop for venture-backed names [119]. A LifeSciVC review of 31 biotech IPOs found no systematic effect [120]. Holders in B-2, B-3, B-4 and the notes are all under water at $11.64, which can reduce selling or not, depending on the fund's mandate.
Dilution still to come. Beyond treasury-method shares, the 2026 Plan has about 3.49 million ungranted shares, with an evergreen of up to 5% of fully diluted a year from FY2027 [1]. New IPO grants, on our Black-Scholes estimate, add about $10 million a year of non-cash expense, about triple the current annualized $5 million of stock comp. The company has not done a Section 382 study on its $157.0 million federal NOL [1].
Index mechanics. IAM listed before the October 30 Russell rank day at about $637 million basic, so we infer it joins the Russell 2000 at the December reconstitution, preliminary list November 13, effective after the close December 11 [121]. Our rough estimate of passive demand, assuming index float of 8.66 million shares excludes locked shares, an unverified assumption, and passive trackers holding 10-20% of float: 0.87 to 1.73 million shares, or $10-20 million at $11.64. That is about 1.9 to 3.8 days of the 454,000-share 10-day average. Order of magnitude only. Float growth after April would prompt more later. NBI looks unavailable until the December 2027 reconstitution on the 2024 methodology, the latest we could read [122].
XV. Valuation
Start with the bridge. These are our reference values.
Method: 47,400,796 basic shares [1]. Diluted adds all 1,231,589 RSUs and PSUs plus treasury-method options and warrants, using S-8 option data split into 286,368 at $12.47 and 3,119,327 at a derived $6.96, plus 14,147 warrants at $3.00 [123][1]. EV subtracts $409.4 million of pro forma June 30 cash, with zero debt because the notes converted. Renaissance's $809 million fully diluted at $16 is close to our $807.5 million [124].
Our adjusted estimate. June 30 pro forma cash is stale. We add the $40 million AbbVie upfront (contracted, timing unconfirmed), add about $9 million of inferred Q3 partner receipts, and subtract our estimate of $35-45 million of gross burn since June 30. That gives cash of about $410-425 million and an adjusted basic EV of about $125-140 million at $11.64, and about $50-65 million at the $10 illustration. It excludes any greenshoe. A full option exercise would add about $16.4 million to market cap and $20.9 million to cash at $11.64, moving unadjusted basic EV to about $138 million. This is our estimate until the first 10-Q replaces it.
Cash per share is $8.64 pro forma, or $9.48 with AbbVie. So at $11.64 the market pays $2.16 to $3.00 a share, before burn, for a Phase 1b drug, two INDs, a platform and partner milestones.
A low EV is not the same as a margin of safety. That cash is planned spend. The company's runway statement implies it is largely consumed by Q4 2028 absent new money. If the plan runs, the equity's value at the end of 2028 is entirely the assets, and investors will likely have been asked for more. A liquidation floor would matter only if spending stopped.
What the EV embeds. Our simple model: peak sales times 3 at a 2030 approval (A), discounted 4 years at 12% (A). We give zero for the platform, IAM217, IAM-C1 and milestones, and we do not net development cost.
Compare base rates. BIO's overall Phase 1 likelihood of approval is 7.9% [125], and a secondary summary gives 5.3% for oncology and 15.9% for biomarker-selected programs [126]. IAM1363 is past first responses on a validated target, so it deserves better than that. At $142 million, you need either a success probability above typical base rates on our base case, or some mix of the bull case, platform and pipeline value. At $16, the market needed 81% on our base case or real platform value. That platform value is what has come out of the stock.
Peers. Prices are Yahoo midday November 6, intraday, against latest SEC balance sheets, mixing June and September, ignoring debt and dilution. Treat these as plus or minus 10-15%.
Source: [127][128]. We infer two things. Against AI peers, at EV of about $1.2-3.2 billion, the AI label earns no premium here. Against Phase 1 and depressed oncology names, at about -$128 million to +$527 million, Iambic's EV is in the normal range. Peers carry different stages, so neither is a proper comp.
Sector context. XBI closed at $148.40 on November 5, up about 22% year to date [29]. A single-source tracker has 2026 US biotech IPOs at a median of about -1.85% at the November 3 close, with 16 of 30 below offer [129]. IAM's -27% is therefore worse than its class.
Deal comps. Seattle Genetics bought Cascadian, and tucatinib, for about $614 million in 2018, with the drug already in pivotal testing [130]. That is the closest HER2 TKI asset comp, and a much later-stage one. GSK's $10.6 billion gross purchase of Nuvalent in June 2026 shows what brain-penetrant TKI platforms can fetch, with value driven by other assets and NVL-330 third in line [102]. Tucatinib inside Seagen's $43 billion price is not a useful comp.
XVI. The bull case, fairly
The drug works in a very hard place. Three confirmed brain PRs after both T-DXd and tucatinib, all still on treatment, 2 beyond six months, with brain PD in only 1 of 22. Nothing is approved in that exact setting.
The confirmed rate has room. Four systemic and one brain PR were pending at the cutoff, with 2 systemic and 3 brain patients on drug with SD approaching PR. A 30% systemic rate is available from pending cases alone.
The dose decision looks real. 960 mg has 5.9% Grade 3 treatment-related events, zero treatment-related discontinuations on the poster, and rash below 10%. The tolerability gap against sevabertinib looks wide, not head to head.
Execution. Management has hit dated milestones, moved to top academic sites, added three pharma deals in eight months and built a pre-registrational team.
The stock has done a lot of the work. EV fell from $349 million to $142 million in three weeks with no adverse filing we can find. About $3.00 a share of premium over pro forma cash buys a clinical asset, two INDs and milestone rights on five deals.
Cash and clean structure. $409.4 million pro forma, plus a contracted $40 million, no debt, no dual class, runway stated to Q4 2028.
Insiders put money in. Ascenta paid $11.5 million in the IPO. Catalio put in $35 million across B-4 and notes. No Form 4 shows a sale. Pre-IPO holders' average cost of $12.45 is above the November 5 close.
Thin float and known demand. Russell inclusion, initiations from four banks and possibly ARK crossing 5%.
Platform as a free option. IAM217 is the real test of NeuralPLexer. The market is paying little for it.
Peak math. Our bull case of $620 million sits a little above Tukysa's $536 million run rate, which is itself growing after HER2CLIMB-05. One trade report cited a Tukysa peak estimate of about $1.5 billion, which we did not verify at the source.
XVII. The bear case, fairly
Confirmed 23% is the number. Post-ADC zongertinib has a 44% confirmed rate in NSCLC, not head to head. No go-forward-dose rate, median duration or PFS has been published. Confirmed fell from 27% to 23% between extracts, with immaturity as the likely reason but not a proven one.
The fastest path is small and unproven. About 16% confirmed in about 19 people, in a population of about 567 a year by our build, worth about $34 million at base.
The primary breast path has a head start against it. BREnnA is randomized and enrolling against the same comparator with Roche's reach. First-line moves of tucatinib and T-DXd also shrink the pool.
Lung is mostly closed. Two approved TKIs in first line, positive Phase 3 data for T-DXd, and Beamion LUNG-2 close to primary completion.
Selectivity did not deliver a clean GI profile. 59% diarrhea, 53% nausea and 49% vomiting at any grade across 960 mg or above. With chronic use and combinations, this matters.
Two pneumonitis cases and unresolved wording. Too few to set a rate, and enough to take a clean ILD claim off the table.
The money does not reach the answer. $75 million does not fund a Phase 3 to readout. Underlying burn rose 25% half on half and rent starts after the runway. We infer a follow-on, likely near a data event and around or after the unlock.
The platform story is unaudited. No Enchant paper, no third-party NeuralPLexer3 benchmark, zero issued platform patents, and the lead drug is not credited to either model.
Partner validation is thin in cash terms. 2.8%, all terminable for convenience, none touching the lead asset, and a Takeda deletion clause limiting compounding.
Supply. Single-source drug substance, no qualified backup, China concentration, WuXi in the vendor base, ahead of a registrational program.
Overhang. About 38 million shares on April 13, 4.06 times the IPO float, and holders who paid more than today's price in seven of nine rows of the entry table.
Where the bear case fails, because a piece written at $16 would be stale.
No evidence of fraud or misconduct appeared. What we found is framing, immaturity and undisclosed detail.
XVIII. What would change the view
These thresholds are our judgment, not sourced standards.
XIX. Dated catalyst calendar
Dates marked "derived" are our arithmetic. Dates marked "custom" reflect practice, not a rule.
No dated IAM1363 readout has been given by the company [8]. The closest things to a real test soon are the 10-Q and initiations.
XX. What we did not get
Courts. CourtListener and Justia were blocked, and we did not run PACER, San Diego Superior Court or Delaware Chancery. The only comfort is the company's statement of no material proceedings. Treat courts as unverified.
SABCS. We did not page through the program. We have no evidence either way on an Iambic abstract.
Greenshoe. No primary source confirms exercise or non-exercise. The 8-K is silent, and no closing release was found. The stock under $16 argues against it now, though the option could have been exercised while the stock traded above $16.
ARK and Duquesne allocations. Not disclosed. Duquesne has no filing naming Iambic. ARK's files show holdings, not source.
Cause of the price drop. We have no explanation for October 19, November 5 or November 6 from any filing or news source. November 6 is also not a close.
CDMO names. Only WuXi AppTec is named, for R&D services. The drug substance maker is unknown.
Exhibits not read in full. We read payment terms and counted redactions in 10.11, 10.12, 10.13, 10.16 and 10.17, not every clause. Exhibits 10.1 to 10.10, 10.14 and 10.15 and the lease amendments were not read. Milestone tables are redacted.
Trial data we wanted. 960 mg-only response, median DoR and PFS, tumor-specific tables, full ILD rates and human brain exposure. The ESMO 2025 poster's filename contains the word draft, and we could not compare it to the presented version.
FDA. FDA's orphan database blocked our query. No meeting minutes or designation letters exist publicly.
Registries. ANZCTR, ChiCTR and Health Canada were not checked. No sites there appear in the US or EU records.
Records. UCC, California and Delaware corporate records, and local hazmat permits were not run. Which round held ISIF's $20 million is unknown.
Counterparty text. We could not read the Takeda 20-F passage or the RevMed Q3 10-Q passage naming Iambic. Partner pipeline archives were not diffed.
People. No oncologist calls, no independent commentary on IAM1363 found, no prescriber survey. Where departed staff went is unknown.
SEC comment letters. None are public yet.
Peer cycle times. We did not gather peer program-start-to-IND data, so the speed claim is not benchmarked against named peers.
XXI. Where this leaves us, and Part 2
The market spent three weeks repricing Iambic from an AI platform with a lead drug to a Phase 1b HER2 pill with cash. That shift matches what the documents support so far: a drug with credible activity in a hard population, a platform with limited outside verification, and a stated plan that costs more than the money on hand. What the price now implies is modest on some measures and still demanding on others. It depends on a confirmed number that was still maturing at the cutoff.
We do not know yet whether 23% becomes 30% or stays 23%. Nobody outside the company does. The next dataset settles more than any writeup.
Part 2 goes one level down. We will take the first 10-Q and rebuild the cash, burn and share count from actuals. We will go through the Part 3 and Part 4 cohorts against what a registrational design needs, including what an FDA reviewer would make of a confirmed rate in the mid-teens after two prior agents. We will read BREnnA's design against Iambic's breast plan, test the KIF18A and CDK2/4 programs against their competitors, and try to close the gaps above: CDMO identity, courts and breast oncologist views. We will also look at how the initiations frame 36 versus 23.
Disclaimer
I am long the names disclosed at the top of this post. This is research synthesis, not investment advice. You should not buy or sell securities based on anything I write. I am not a registered investment advisor. I do not owe you a fiduciary duty. My conclusions could be wrong in ways I have not anticipated. Financial projections are model outputs based on publicly available data. They are not guarantees. Microcaps are illiquid and can lose 50% or more in a single session. Do your own due diligence. Verify every number against primary sources.
References
Iambic Therapeutics 424B4, final prospectus dated Oct 14, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000119312526422865/iambic_inception_424b4.htm
DRS/A, Aug 28, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000119312526374173/filename1.htm
crt.sh certificate transparency log, entos.ai: https://crt.sh/?q=%25.entos.ai
ESMO 2026 poster, FPN 1072P: https://cdn.prod.website-files.com/64d1614b6114857407c79035/6adf7d2c6814a594b84a6cee_IAM1363-01%20ESMO%202026%20Poster_2026-10-16%20Final%20(1).pdf
Iambic IAM217 IND release, Oct 5, 2026: https://www.iambic.ai/post/iambic-submits-ind-for-a-potentially-brain-penetrant-kif18a-inhibitor-iam217-its-second-wholly-owned-ai-discovered-candidate
ClinicalTrials.gov NCT06253871, current record: https://clinicaltrials.gov/api/v2/studies/NCT06253871
AACR 2025 KIF18A poster: https://cdn.prod.website-files.com/64c03f218d60985b40eeeae5/68c083ba9b3cccd29b8d1562_iambic-a-novel-KIF18A.pdf
Iambic ESMO 2026 press release: https://www.iambic.ai/post/iambic-presents-updated-iam1363-clinical-data-at-esmo-2026
S-1, Sept 21, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-20260921.htm
Pharmacally summary of ESMO 2026 data: https://pharmacally.com/?p=26906
DRS, July 24, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000119312526315779/filename1.htm
ESMO 2025 poster, FPN 938P: https://cdn.prod.website-files.com/64c03f218d60985b40eeeae5/68f101ea0836145f857dd7b7_IAM1363-01%20ESMO%202025%20Poster_Author%20Mgmt%20Review_Draft%202025-09-30.pdf
PubMed search, IAM1363: https://pubmed.ncbi.nlm.nih.gov/?term=IAM1363
ClinicalTrials.gov NCT06253871, version history: https://clinicaltrials.gov/study/NCT06253871?tab=history
START trial page, IAM1363-01: https://hopehub.startresearch.com/trials/IAM1363-01
EU CTIS record 2024-519188-17-00: https://euclinicaltrials.eu/ctis-public-api/retrieve/2024-519188-17-00
Iambic HER2 IND release, Dec 2023: https://www.iambic.ai/post/her2-ind
Iambic pipeline page, archived May 2025: https://web.archive.org/web/20250521091421/https://www.iambic.ai/pipeline
WCLC 2025 poster: https://cdn.prod.website-files.com/64d1614b6114857407c79035/68c04f6931e5573142ab8e27_HER2_2025%20WCLC%2020250902.pdf
Iambic pipeline page, archived Nov 2025: https://web.archive.org/web/20251118201227/https://www.iambic.ai/pipeline
EMA orphan designations file: https://www.ema.europa.eu/en/documents/report/medicines-output-orphan_designations-json-report_en.json
Form 4, Catalio, Oct 2026: https://www.sec.gov/Archives/edgar/data/1997038/000195812526000011/primary_doc.xml
Form 4, Evan Rachlin and Ascenta, Oct 19, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000215167826000007/form4-10192026_111032.xml
Form D, April 8, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000199703826000001/primary_doc.xml
EDGAR submissions feed for CIK 1997038: https://data.sec.gov/submissions/CIK0001997038.json
S-1/A, Oct 8, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000119312526417416/iam-20261008.htm
Nasdaq historical price data, IAM: https://api.nasdaq.com/api/quote/IAM/historical?assetclass=stocks&fromdate=2026-10-14&limit=30&todate=2026-11-06
Nasdaq quote data, IAM, Nov 6 intraday: https://api.nasdaq.com/api/quote/IAM/info?assetclass=stocks
Yahoo Finance, XBI: https://finance.yahoo.com/quote/XBI/
Exhibit 10.16, 3545 Cray Court lease: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-ex10_16.htm
Iambic financing release, Nov 2025: https://www.iambic.ai/post/iambic-raises-over-100-million
ISIF portfolio: https://www.isif.ie/portfolio
ISIF newsletter, April 2026: https://www.isif.ie/isif-newsletter-3
Business Barometer, Iambic Therapeutics IRE Limited: https://www.businessbarometer.ie/Risk-Management/Iambic-Therapeutics-Ire-Limited-814819
Exhibit 21.1, subsidiaries: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-ex21_1.htm
NIH RePORTER API: https://api.reporter.nih.gov/v2/projects/search
USAspending search API: https://api.usaspending.gov/api/v2/search/spending_by_award/
Senate LDA filings, client Iambic: https://lda.senate.gov/api/v1/filings/?client_name=Iambic
CMS Open Payments: https://openpaymentsdata.cms.gov/
US12145948B2: https://patents.google.com/patent/US12145948B2/en
NeuralPLexer, Nature Machine Intelligence, 2024: https://doi.org/10.1038/s42256-024-00792-z
OpenAlex affiliation query: https://api.openalex.org/works?filter=raw_affiliation_strings.search:%22Iambic%20Therapeutics%22&per-page=100
NeuralPLexer3, arXiv 2412.10743: https://arxiv.org/abs/2412.10743
Iambic NeuralPLexer3 preview post: https://www.iambic.ai/post/np3-preview
PoseBench, arXiv 2405.14108: https://arxiv.org/html/2405.14108v6
Iambic Enchant whitepaper, Oct 2024: https://www.iambic.ai/post/enchant
Iambic Enchant v2 post, May 2025: https://www.iambic.ai/post/enchantv2
Iambic Enchant v3 release, archived: https://web.archive.org/web/20261016104106/https://www.iambic.ai/post/iambic-launches-enchant-v3---molecular-superintelligence-designed-to-advance-end-to-end-drug-discovery-development
Research Square preprint on probability of success prediction, July 2026: https://doi.org/10.21203/rs.3.rs-9416583/v1
Iambic about page: https://www.iambic.ai/about
Iambic and Lambda release, June 2025: https://www.businesswire.com/news/home/20250626985107/en/Iambic-Taps-Lambda-for-Increased-AI-Compute-as-it-Develops-the-Next-Generation-of-Enchant-a-Breakthrough-AI-Molecular-Property-Prediction-Model-for-Drug-Discovery
Exhibit 10.12, Takeda agreement: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-ex10_12.htm
Exhibit 10.13, Lundbeck agreement: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-ex10_13.htm
Exhibit 10.17, AbbVie agreement: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-ex10_17.htm
BioPharma Dive interview with Tom Miller, Oct 2026: https://www.biopharmadive.com/news/Iambic-ipo-tom-miller-ai-drug-development-discovery/833238/
Revolution Medicines FY2025 10-K: https://www.sec.gov/Archives/edgar/data/1628171/000119312526071563/rvmd-20251231.htm
AbbVie Q3 2026 press release: https://www.sec.gov/Archives/edgar/data/1551152/000155115226000033/abbv-20260930xexhibit991.htm
Takeda FY2025 20-F: https://www.sec.gov/Archives/edgar/data/1395064/000139506426000177/tak-20260331.htm
Holland & Knight on WuXi AppTec and 1260H, June 2026: https://www.hklaw.com/en/insights/publications/2026/06/wuxi-apptec-added-to-dows-1260h-list
Greenberg Traurig on BIOSECURE and the WuXi injunction, Sept 2026: https://www.gtlaw.com/en/insights/2026/9/biosecure-act-update-court-enjoins-wuxi-apptecs-1260h-designation-but-strategic-and-compliance-clocks-keep-running
Iambic job board, Ashby public API: https://api.ashbyhq.com/posting-api/job-board/iambic-therapeutics
Iambic job board, archived Sept 2026: https://web.archive.org/web/20260904113014/https://jobs.ashbyhq.com/iambic-therapeutics
California EDD WARN report: https://edd.ca.gov/siteassets/files/jobs_and_training/warn/warn_report1.xlsx
Companies House, company 12789012: https://find-and-update.company-information.service.gov.uk/company/12789012
Iambic Therapeutics UK Ltd FY2025 accounts: https://find-and-update.company-information.service.gov.uk/company/12789012/filing-history/MzUxNzEzMjEzNmFkaXF6a2N4/document?format=xhtml&download=1
EPA ECHO, 5627 Oberlin Dr: https://echodata.epa.gov/echo/dfr_rest_services.get_dfr?output=JSON&p_id=110071676981
EPA ECHO, 3545 Cray Ct: https://echodata.epa.gov/echo/dfr_rest_services.get_dfr?output=JSON&p_id=110072168486
Exhibit 16.1, Ernst & Young letter: https://www.sec.gov/Archives/edgar/data/1997038/000119312526396704/iam-ex16_1.htm
Iambic CMO release, archived Feb 2024: https://web.archive.org/web/20240225203407/https://www.iambic.ai/post/iambic-therapeutics-expands-leadership-team-with-the-appointment-of-neil-josephson-md-as-chief-medical-officer
Form 4, Tom Miller, Oct 16, 2026: https://www.sec.gov/Archives/edgar/data/1997038/000203492126000007/form4-10162026_081045.xml
Form 3, Tom Miller: https://www.sec.gov/Archives/edgar/data/1997038/000203492126000005/form3-10142026_101008.xml
Iambic homepage, archived Oct 2023: https://web.archive.org/web/20231005110722/https://www.iambic.ai/
Iambic homepage, archived May 2026: https://web.archive.org/web/20260508110233/https://www.iambic.ai/
Iambic homepage, archived Oct 2026: https://web.archive.org/web/20261025094618/https://www.iambic.ai/
Iambic pipeline page, archived Oct 2023: https://web.archive.org/web/20231017081303/https://www.iambic.ai/pipeline
Iambic pipeline page, archived Oct 2026: https://web.archive.org/web/20261025094630/https://www.iambic.ai/pipeline
Iambic partnerships page, archived Feb 2024: https://web.archive.org/web/20240225223701/https://www.iambic.ai/partnerships
Iambic partners page, archived Oct 2026: https://web.archive.org/web/20261016104106/https://www.iambic.ai/partners
Endpoints headline archived by Iambic, Oct 2023: https://web.archive.org/web/20231207090353/https://www.iambic.ai/post/endpoints-news-ai-driven-biotech-iambic-snags-100m-to-enter-the-clinic-next-year-in-her2-field
Iambic Series B release, Oct 2023: https://www.iambic.ai/post/series-b
Iambic CFO release, Jan 2025, archived: https://web.archive.org/web/20250122220726/https://www.iambic.ai/post/iambic-therapeutics-appoints-michael-secora-ph-d-as-chief-corporate-development-officer-and-chief-financial-officer
Iambic ESMO 2025 press release: https://www.iambic.ai/post/esmo-iam1363
Hernexeos FDA label, revised 2026: https://www.accessdata.fda.gov/drugsatfda_docs/label/2026/219042s006lbl.pdf
FDA, zongertinib first-line accelerated approval: https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-zongertinib-unresectable-or-metastatic-non-squamous-non-small-cell
FDA, sevabertinib first-line accelerated approval: https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-sevabertinib-locally-advanced-or-metastatic-non-squamous-non-small
Hyrnuo FDA label, 2025: https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/219972s000lblCorrected.pdf
ecancer on DESTINY-Lung04, WCLC 2026: https://ecancer.org/en/news/28900-wclc-2026-phase-3-destiny-lung04-trial-shows-first-line-trastuzumab-deruxtecan-significantly-improves-progression-free-survival-in-her2-mutant-nsclc
Targeted Oncology on sequencing after zongertinib: https://www.targetedonc.com/view/second-line-sequencing-after-progression-on-zongertinib-adc-tki-switch-or-chemotherapy
OncLive panel on CNS control in HER2-mutated NSCLC, Sept 2026: https://www.onclive.com/view/cns-control-and-daily-tolerability-guide-treatment-sequencing-in-her2-mutated-nsclc
HER2CLIMB, NEJM: https://pubmed.ncbi.nlm.nih.gov/31825569/
HER2CLIMB brain metastases, JCO: https://pubmed.ncbi.nlm.nih.gov/32468955/
DESTINY-Breast03, NEJM: https://pubmed.ncbi.nlm.nih.gov/35320644/
DESTINY-Breast12, Daiichi release: https://www.businesswire.com/news/home/20240912472756/en
DESTINY-Breast09 brain metastases subgroup, OncLive: https://www.onclive.com/view/t-dxd-plus-pertuzumab-improves-outcomes-in-her2-breast-cancer-with-brain-metastases
ASCO Post on HER2CLIMB-05 approval, Oct 2026: https://ascopost.com/news/october-2026/her2-positive-breast-cancer-fda-approves-triplet-for-maintenance-therapy/
ClinicalTrials.gov NCT07413939, BREnnA: https://clinicaltrials.gov/study/NCT07413939
Roche trial page, RO7771950 vs tucatinib: https://forpatients.roche.com/en/trials/cancer/bc/ro7771950-versus-tucatinib-in-combination-with-trastuzu-54968.html
Dana-Farber blog, Sept 2026: https://blog.dana-farber.org/insight/2026/09/patient-breast-cancer-brain-metastases-research-advances/
ClinicalTrials.gov NCT06151574, Beamion LUNG-2: https://clinicaltrials.gov/study/NCT06151574
ClinicalTrials.gov NCT06452277, SOHO-02: https://clinicaltrials.gov/study/NCT06452277
ClinicalTrials.gov NCT06324357, Beamion BCGC-1: https://clinicaltrials.gov/study/NCT06324357
AJ Bell on GSK and Nuvalent, June 2026: https://www.ajbell.co.uk/news/articles/update-gsk-buys-cancer-drug-developer-nuvalent-106-billion
Fierce Biotech on ELVN-002: https://www.fiercebiotech.com/biotech/enliven-calls-time-phase-1-her2-solid-tumor-program
Pfizer Tukysa price disclosure, Jan 2025: https://cdn.pfizer.com/pfizercom/products/coprescribers/shortforms/TUKYSACOPriceDisclosureForm01012025.pdf
Pfizer Q3 2026 earnings release: https://s206.q4cdn.com/795948973/files/doc_financials/2026/q3/Q3-2026-PFE-Earnings-Release-FINAL.pdf
AstraZeneca 6-K, 9M 2026: https://www.sec.gov/Archives/edgar/data/0000901832/000165495426009256/a0835b.htm
SEER lung: https://seer.cancer.gov/statfacts/html/lungb.html
SEER breast subtypes: https://seer.cancer.gov/statfacts/html/breast-subtypes.html
SEER stomach: https://seer.cancer.gov/statfacts/html/stomach.html
SEER esophagus: https://seer.cancer.gov/statfacts/html/esoph.html
Real-world HER2-mutant NSCLC, Cancer Medicine: https://pmc.ncbi.nlm.nih.gov/articles/PMC11653943/
Pharmacy Times on metastatic breast cancer incidence: https://www.pharmacytimes.com/view/metastatic-breast-cancer-prevalence-continues-to-grow
OncLive on real-world brain metastasis incidence in HER2+ breast cancer: https://www.OncLive.com/view/real-world-analysis-sheds-light-on-brain-metastases-incidence-in-her2-breast-cancer
GxP News on Hyrnuo forecasts: https://gxpnews.net/en/2026/09/bayer-strengthens-market-position-with-expanded-hyrnuo-indication/
8-K Exhibit 3.1, charter: https://www.sec.gov/Archives/edgar/data/1997038/000119312526424242/iam-ex3_1.htm
ARKG daily holdings file, Nov 6, 2026: https://assets.ark-funds.com/fund-documents/funds-etf-csv/ARK_GENOMIC_REVOLUTION_ETF_ARKG_HOLDINGS.csv
ARK Venture Fund N-CSR, schedule of investments: https://www.sec.gov/Archives/edgar/data/1905088/000121390026107607/ea0304362-01_ncsr.htm
ARK Venture Fund N-PORT, July 31, 2026: https://www.sec.gov/Archives/edgar/data/1905088/000094040026036464/primary_doc.xml
Field and Hanka, Journal of Finance, 2001: https://ideas.repec.org/a/bla/jfinan/v56y2001i2p471-500.html
LifeSciVC on post-IPO lock-up expiry: https://lifescivc.com/2018/10/the-fear-of-the-flood-post-ipo-lockup-expiry/
LSEG, December 2026 Russell reconstitution schedule: https://www.lseg.com/en/media-centre/press-releases/ftse-russell/2026/ftse-russell-announces-december-2026-russell-us-indexes-reconstitution-schedule
Nasdaq Biotechnology Index methodology: https://indexes.nasdaqomx.com/docs/methodology_NBI.pdf
S-8 fee table, Exhibit 107: https://www.sec.gov/Archives/edgar/data/1997038/000119312526421986/iam-exfiling_fees.htm
Renaissance Capital on IPO terms: https://www.renaissancecapital.com/IPO-Center/News/122117/small-molecule-biotech-iambic-therapeutics-sets-terms-for-150-million-ipo
BIO clinical development success rates 2011-2020: https://www.bio.org/clinical-development-success-rates-and-contributing-factors-2011-2020
Secondary summary of BIO success rates: https://www.cn-healthcare.com/articlewm/20210508/content-1217774.html
Yahoo Finance quote data, Nov 6, 2026 midday: https://finance.yahoo.com/quote/IAM
SEC XBRL company facts API: https://www.sec.gov/edgar/sec-api-documentation
BioBucks 2026 biotech IPO tracker: https://www.biobucks.co/biotech-ipo-tracker-2026
Fierce Biotech on Seattle Genetics and Cascadian: https://www.fiercebiotech.com/biotech/seattle-614m-buyout-cascadian-and-its-breast-cancer-program
FINRA Rule 2241: https://www.finra.org/rules-guidance/rulebooks/finra-rules/2241
MarketBeat, IAM: https://www.marketbeat.com/stocks/NASDAQ/IAM/
SABCS 2026 abstract dates: https://sabcs.org/?p=5319
SEC Rule 13a-13: https://www.law.cornell.edu/cfr/text/17/240.13a-13
AACR Annual Meeting 2027: https://www.aacr.org/meeting/aacr-annual-meeting-2027/