The bull case rests on one technical argument
The most AI-specific claim in this week's Palantir coverage is about architecture. An analyst, Ryan Canady, upgraded Palantir to Buy after concluding that customer-built AI applications can remain complementary to Palantir's platforms rather than displace its core offering.1 That addresses a live worry for practitioners: if enterprises can assemble their own AI applications, a platform vendor's layer might shrink. The upgrade says it need not.
A second analyst, Bernard Zambonin, remains constructive. He emphasizes customer expansion and net dollar retention as central to the bull case, and acknowledges that the elevated forward valuation makes continued execution essential.1 Net dollar retention measures whether existing customers spend more over time. Both calls depend on customers adopting more of the platform, not just signing up.
Our narrative tracking also flags a Goldman Sachs upgrade in this cluster of calls. The dossier gives no detail on that upgrade, so we do not characterise it.
What the October partnerships actually are
Three October announcements show where Palantir is trying to deploy AI. Each is described below only as far as the record goes.
On October 14, Customers Bank announced a collaboration with Palantir to develop an AI-oriented banking operating model. Its own forward-looking statements highlight implementation risks, reliability concerns and contractual flexibility.2 That is a bank telling investors the project could go wrong, which is useful context for a partnership headline.
The same day, Surf Air Mobility signed Flying Zebra as its fifth commercial OperatorOS customer. The software is powered by Palantir, and Surf Air receives a percentage of revenue from trips managed through the platform.3
On October 13, Ondas and Palantir announced Regna AI. It is a standalone private company meant to help governments, businesses and institutions develop and deploy AI models in sovereign environments. Palantir is expected to provide model-training infrastructure and engineering.4 A sovereign environment keeps data and models inside a customer's own control. That matters for governments and regulated firms that cannot send data to a shared cloud.
The market's reaction: a 7% drop within days of the highs
Good news did not shield the stock. Palantir fell more than 7% on Friday, October 16, closing at $186.03 after a $14.47 decline. It had finished lower in five consecutive sessions, coming off a 52-week high of $212.94 set early that week. The drop came during a broader technology selloff.1
Zambonin's caveat, that the elevated forward valuation makes continued execution essential, reads differently next to those numbers.1 The bulls and the market are both pricing in very strong execution. One day's move does not show the thesis is wrong, but it shows how little room a priced-for-growth stock has when sentiment turns.
A note on the source: the article comes from NewsEOD, whose claims have held up 57% of the time across 881 we checked.1 That is a measured figure for the outlet, not for this article. Treat the price figures as reported, not independently verified.
The other half of the divergence: traditional defense
The contrast in this story is with companies that sell to government in more traditional ways. J.P. Morgan's lead aerospace and defense analyst, Seth M. Seifman, wrote on October 8 that most companies should deliver solid third-quarter results and express confidence in underlying demand. He added that good earnings may not reverse the sector's stock-market weakness.5 The report cited slowing airline traffic growth, uncertainty over U.S. military spending and lingering supply-chain constraints.5
J.P. Morgan downgraded Leidos. Seifman's reasoning was that Leidos' relatively low valuation is not enough to justify an Overweight rating when other aerospace and defense stocks also offer substantial potential returns.5 The dossier also records a competitive link, with BigBear.ai Holdings, Inc. listed as competing with Leidos Holdings, Inc.6
Palantir sells to government and commercial customers alike, a point the October 16 article makes itself.1 We do not have data in the dossier splitting its revenue between the two, so we cannot say how much of its story depends on defense budgets. What we can say is that Palantir's analysts are talking about commercial AI adoption, while the defense analyst is talking about budget uncertainty.
Political event risk
The November 3 congressional midterm elections could change congressional control and constrain military spending growth. Some of the uncertainty may dissipate after that date.7 That is the clearest scheduled catalyst for the defense side of this story.
Other AI-adjacent names: expensive, and contested
The market is also paying high prices for AI-adjacent stories, and the analyst views are far from uniform.
On quantum computing, Melissa Tucker's analysis says D-Wave (QBTS) trades at a stretched 171x price-to-sales multiple versus peers. It projects 2026 revenue near $35 million and a fair value of $9.50 per share, implying roughly 40% downside. Her reasons were a delayed technology roadmap and D-Wave's reliance on SkyWare, amid IonQ's pending acquisition of that supplier.8 Another analyst, Marc Gerstein, sees D-Wave's dual-platform approach, spanning annealing and gate-model quantum computing, as a competitive differentiator and strategic advantage over single-method peers.8 The stock closed at $17.64 after falling $1.88, and was down 25% over the past month.8 Two analysts, two opposite readings of the same company.
On AI power demand, Nicholas Amicucci argues that Bloom Energy's ability to provide reliable, dispatchable power to a volatile demand profile differentiates it from competitors.9 The same piece notes Bloom's inclusion in the Russell 1000 index and an expansion of its AI infrastructure partners.9 On nuclear, Bank of America's Rinny Singh reiterated a buy rating on NuScale (SMR) with a $12 price target, implying roughly 24% upside, though the shares have fallen nearly 40% this year.10 That Motley Fool article comes from a source where only 22% of 18 checked claims held up, a small sample but a weak result, so we weight it lightly.10
What our verified data can and cannot say
Our fidelity-checked database holds no Palantir financials for this piece. The verified figures we have concern Goldman Sachs, which appears in the narrative as an upgrader. They are SEC-filing figures for the bank itself, not evidence about Palantir. Goldman's full-year earnings per share were $22.87 in fiscal 2023, $40.54 in 2024 and $51.32 in 2025.11 Its capital expenditure was $2.316 billion in 2023, and $2.064 billion in 2025.12 The bank's earnings per share more than doubled over that span while capital spending slipped. This is background on one bank, not a signal about Palantir.
To be plain about the gap: the analyst calls on Palantir are opinions. Nothing in our verified data confirms customer expansion, net dollar retention or the size of the new partnerships. Readers should treat those as claims to be tested at the next earnings report, not as established facts.
What to watch
- November 3: the midterm elections, which could change congressional control and constrain military spending growth.7
- Third-quarter defense results: Seifman expects solid numbers but warns they may not lift the stocks.5
- Whether Palantir recovers its highs: it closed at $186.03 against a 52-week high of $212.94.1
- The Customers Bank project: the bank's own risk language about implementation and reliability is the benchmark for judging results.2