Alex Karp did not use the words "beat and raise" in Monday evening's earnings report, which saw Palantir beat expectations and raise its outlook.
First, he used "otherworldly" to describe his company's performance, and then the mercurial tech billionaire and CEO told CNBC that investors should "forget consensus", presumably given his company's current blistering rate of growth.
The numbers gave him room for such confidence.
Palantir, Karp's 'big-data' software contractor to a customer base comprising governments, banks and many other forms of international institutions, on Monday revealed a 93% jump in quarterly revenue, reaching $1.935 billion. It blew past market consensus pitched between $1.80–1.81 billion.
In the United States, Palantir's revenue growth marked an even higher rate of 149%, to $764 million. At the same time, earnings were marked at $0.41 a share, versus a $0.35 per share forecast. And, the firm lifted full-year revenue guidance to $8.150–8.158 billion, from $7.65–7.66 billion.
With trademark modesty and formality, the investor relations copy accompanying the figures spelt out the company's sentiment. It was "crushing consensus".
In after-hours trade, Palantir stock soared some 14% higher, changing hands at $143.4.
Now, we could keep crunching the numbers, or perhaps there's the option of taking Karp's lead and forget the analysis of metrics and instead, we could hang on his words.
Karp's pitch lands squarely on "AI sovereignty"
Sovereign AI is one of those buzzphrases that's been thrown around a lot in recent years. Generally, and in most cases, it's a grandiose way of saying that automations within business organisations use their own protected, private data sets (rather than sending them into OpenAI or Anthropic data centres), and/or systems using proprietary AI models. Simply put, such systems give data owners dominion and control over how their data is used, stored and mutated by the systems that consume and process it.
For Karp, though, with Palantir's governmental clients across healthcare, police, immigration and military departments and agencies, the phrase both means the more literal big-S sovereign and as well as the smaller-verbish reference.
On both sides of this particular coin, Karp's message was as clear as it was (predictably) bold.
Palantir clients trust the data firm to provide maximal control without exposure, according to Karp, who claims his company is the only one able to "transform tokens into actual economic value".
Karp says "demand AI sovereignty has now been unleashed", and that the "sovereign AI revolution" makes him very optimistic about the future.
His separate Letter to Shareholders extols further.
A global movement is emerging.
The revolution for independence and AI sovereignty is now well underway.
Every organization in the world is awakening to the risks of handing the creators of the language models the keys to their institutions, of letting the models loose within their homes.
The demand from our partners is clear.
The 'language labs' behind LLMs are structurally designed to capture control over proprietary data, business intelligence, and 'alpha', according to Karp, who claims that having already ingested the entire written work of our civilisation, these labs now have their sights set on global industry.
But the man who runs the software company that handles the personal data of tens of millions of citizens around the world is here to reassure, as he explained that Palantir is one of the few companies in the technology industry that is fully aligned with its customers.
Flicking back through the letter to shareholders, skipping past distinctly separate excerpts which on one hand describe an "astonishing level" of profits and commercial growth at "unrelenting and breakneck speed", and on the other talks up the "Marxist overtones and undertones to our business" . . . Karp lands what's likely the intended core insight.
Our customers have declined to become vassal states of the language labs. And the market is now shifting dramatically underfoot.
The veracity of such a comment likely depends upon which customers are being talked about and what kinds of integrations are being described; given that several major governments now have active partnerships directly with both OpenAI, Anthropic and Google.
Big tech CEOs set out their stalls
Realistically, this has been a reporting season that's had one thematic constant from one IR splash to the next.
Against a market backdrop of scepticism and scrutiny, with market pundits' sharpened gaze on anything bubble-ish, it has been a summer reporting season that's seen tech billionaires setting out their stalls.
For Meta's Mark Zuckerberg is all about helping humanity, with the Facebook owner's message that "the future is for everyone" . . . (especially potential customers who may want to rent the social media company's excess compute), and, Apple's arguably strong AI pitch, which more or less boils down to 'we're AI winners because we're not burning capital trying to compete, and instead we'll just sell a ton more iPhones and Macbooks".
Let's not forget OpenAI's IR contribution, which, having paused IPO efforts, amounted to some reportage on a seemingly bullish internal employee-facing comms which suggested July's month of trading was stronger than its prior quarter.
And, Tuesday will bring both noise and insight, when Elon Musk's SpaceX - now the stock market-listed parent of xAI and 'Grok' - delivers its first set of financials since its record-breaking float. SpaceX started the week close to 30% beneath its IPO, so Tuesday's earnings release will be closely watched - and so will be Thursday's trading, as the first set of share 'lock-ups' expire (allowing certain pre-IPO holders to cash out some shares).
Microsoft's IR message was that the heavy capex deployed so far is now beginning to pay off with tangible results, or, as Satya Nadella emphasised, the Windows-owner is seeing an improving "cost-to-outcome curve".
Whilst Google's enthusiasm for its AI Search Overviews appeared unabating in its earth-scorching though the independent internet, which is increasingly starved of click-through-traffic but continues to provide substance to assimilated search answers. From the outside-looking-in, it's unclear how Google plans to sustain the revenue-beast that is its internet-wide ad-business, whilst cutting off clicks to websites, but the message to investors was that AI isn't hurting 'the core business'.
Whichever direction one looks, the upshot is that, clearly, there's more than one AI story to tell - and more than one AI pitch to sell. And, for Palantir, it's about 'orchestration' and the control (and leverage) of proprietary data.
What does the market think?
Well, the stock rallied - adding 14% after hours - and continued to point higher ahead of Tuesday's open.
But, prior to the earnings release, Palantir began the week down roughly 30% year-to-date, and about 40% off its November highs.
Taken at face value, there's no question from the company's IR that there's a shortage of bullishness in-house.
Karp's pitch is not about model capability, web-artefact gizmos, or token usage. It is about control. His company enables "AI sovereignty", allowing large organisations to keep their own competitive advantage by keeping them out of somebody else's training set.
It's an echo to Karp's previous advocacy for open-weight models, and his recent warnings that America should be developing open models that must match Chinese ones (so the US, and by extension the west, isn't overly reliant on the frontier labs).
Whether Karp's positioning and sovereignty pitch is a durable story for clients and markets will be an open question for a while.
Analysts in the sector reckon that the 'orchestration layer', proprietary software harnesses that can switch between LLM models with minimal disruption, are the defensible asset - and, by extension, doing orchestration with hardened data security extends that moat.
More broadly, judging by market punditry across the wires and the TV-studio talking heads, there's the view that the AI sector faces a deceleration of these rampant growth rates. For context, decelerative still means a 'slower' rate of growth at around 50% over twelve months.
Ever-bullish Karp, however, reckons it "looks like this is going to go on for at least another 18 months."
