The short version, and an honest caveat
Two things happened in the same fortnight. Several analysts became more positive on Palantir, the US software and AI company that serves government and commercial customers.1 Meanwhile J.P. Morgan's lead aerospace and defence analyst described a challenging earnings season for the sector and downgraded Leidos, a US defence contractor.2 The two stories pull in opposite directions, and they show how differently the market treats an AI software platform and a defence contractor.
For a European business reader there is a limit to what we can say. Our verified material contains no EU regulation, no European company results and no European market data on this story. We will not invent a European angle. The one thread that touches European concerns is sovereign AI, covered below. We flag where the evidence stops.
What happened to Palantir's shares
On Friday 16 October 2026 Palantir fell more than 7% to close at $186.03, a drop of $14.47. It was the fifth lower close in a row, following a 52-week high of $212.94 set earlier that week. The fall came during a broader technology selloff that weighed on the Nasdaq-100.1 If you hold a global or US technology index fund, you may own some Palantir indirectly. The dossier does not give its weighting in any index, so we cannot say how much.
The bullish calls arrived in the same stretch. Bernard Zambonin stays constructive on Palantir. He points to customer expansion and net dollar retention (how much more existing customers spend year on year) as central to the bull case. He also acknowledges that the elevated forward valuation makes continued execution essential.1 In plain terms, the shares are priced for things to keep going well, and the analyst says so.
Ryan Canady upgraded Palantir to Buy. He concluded that customer-built AI applications can remain complementary to Palantir's platforms rather than displace its core offering.1 That addresses a real investor worry, that customers might build their own tools instead of paying for Palantir's. It is a judgement about the future, not a measured result.
The partnership run
Palantir's announcements in the same days point the same way, towards ecosystem deals:
- On 14 October, 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.3
- On 14 October, Surf Air Mobility signed Flying Zebra as its fifth commercial OperatorOS customer, with software powered by Palantir. Surf Air receives a percentage of revenue from trips managed through the platform.4
- On 13 October, Ondas and Palantir announced Regna AI, a standalone private company to help governments, businesses and institutions develop and deploy AI models in sovereign environments. Palantir is expected to provide model-training infrastructure and engineering.5
The Customers Bank wording is a useful reminder. A partnership announcement is not revenue, and the bank itself lists reliability and implementation risk as things that could go wrong.3
Regna AI is the closest this story gets to European concerns. Keeping AI models inside a government's or institution's own controlled environment is a theme European buyers care about. However, the dossier does not say that any European customer is involved, and we make no claim that one is.5
The defence side: a different mood
In an 8 October research report, J.P. Morgan's lead aerospace and defence analyst, Seth M. Seifman, said most companies should deliver solid quarterly results and express confidence in underlying demand. But encouraging earnings may not be enough to reverse the sector's recent stock-market weakness.2 The report cites slowing airline traffic growth, uncertainty over US military spending and lingering supply-chain constraints as weights on investor sentiment.2
On Leidos, Seifman's view is that its relatively low valuation is insufficient to justify an Overweight rating when other aerospace and defence stocks also offer substantial potential returns.2 That is a relative call. It says other defence shares look at least as attractive, not that Leidos is a bad business. Seifman also expects Howmet's 2026 earnings estimates to rise, with management confident about growth in aerospace and industrial gas turbines.2 So the sour mood is not uniform across the sector.
Political timing adds to the uncertainty. The US congressional midterm elections on 3 November 2026 could change congressional control and constrain military spending growth. Some uncertainty may dissipate after that date.6 Our entity data also records BigBear.ai as a competitor of Leidos.7 We have no further data on that relationship, so we draw no conclusion from it.
Is this really a contrast?
The framing of AI momentum against defence gloom is neat, and we treat it with caution. Palantir sells to both government and commercial customers,1 so it is not simply on the other side of the defence trade. The dossier gives no Palantir revenue split, no defence exposure figure and no Palantir financial statement. Without those we cannot say how much a weak defence tape matters for Palantir. This is an observation about timing and mood, not a measured link.
The Palantir selloff also complicates the bullish story. The shares fell about 7% in a day on a broad tech decline, after analysts had turned more positive.1 Good commentary and a rising share price are different things.
How far to trust the sources
Every claim above comes from NewsEOD articles. Our own measurement of that source puts its reliability at 58% of 927 checked claims holding up.1,2 That is a meaningful miss rate, and the reason we attribute each view to a named analyst rather than stating it as fact. The analyst quotes are what the articles report those analysts as saying. We have not seen the underlying research notes.
The verified, SEC-filing-backed figures in our dossier concern Goldman Sachs, not Palantir or Leidos, so we have not used them here. Nothing in this piece is a checked Palantir financial figure.
What to watch
- 3 November: the US midterm elections. J.P. Morgan flagged them as a source of defence budget uncertainty that may ease afterwards.6
- Palantir's execution: Zambonin ties the bull case to customer expansion and net dollar retention.1 Any updated figures on these would test the case directly.
- Regna AI details: who the customers are, and whether any are European, would show whether the sovereign-AI idea reaches this audience.5
- Defence earnings: whether solid results actually lift the sector, as Seifman doubts they will.2
