Raspberry Pi Stock Analysis: What the Latest RPI Results Mean for Investors

Raspberry Pi Holdings reported first-half revenue of USD 256.9m on 24 September 2026, up 90% on the USD 135.5m booked in the same six months of 2025, and told the market that full-year adjusted EBITDA should land ahead of consensus. The shares closed up 16.04% at 732.81p, and have more than doubled so far in 2026. For anyone running a Raspberry Pi stock analysis off the back of that print, the number that matters is not the 90%. It is the 17%.

That 17% is the growth in units shipped: 4.2m boards against 3.6m a year earlier. The rest of the revenue step-up is price. Average selling price rose 42% to USD 65.90 from USD 46.40, and the company has been explicit that the increase reflects elevated memory costs passed through to customers. Adjusted EBITDA more than doubled to USD 40.3m from USD 19.4m, profit before tax rose 216% to USD 19.6m and adjusted basic earnings per share rose 192% to 13.90 cents. Those are records. They are also, on management's own wording, partly the product of exceptional unit economics that have already moderated. So the honest read on Raspberry Pi is a business with a real demand story sitting underneath a margin number that will not repeat in the same shape.

What actually happened

The half to the end of June 2026 was the strongest six months of Raspberry Pi's short life as a public company. Revenue of USD 256.9m was a 90% advance. Adjusted EBITDA of USD 40.3m was more than double the USD 19.4m of the prior half, and profit before tax of USD 19.6m was up 216%. Adjusted basic EPS of 13.90 cents was up 192%. The company shipped 4.2m units, a 17% increase, and carried a backlog of 2.6m units into the second half.

Two things drove the gap between a 17% volume increase and a 90% revenue increase. The first is mix and price. The 42% rise in average selling price to USD 65.90 is overwhelmingly a memory story: DRAM pricing has moved sharply, and Raspberry Pi has raised board prices to reflect it. The second is that boards are no longer the whole company. Silicon, compute modules and the higher-value industrial lines carry different economics from the single-board computers that made the brand, and they are growing into the revenue base.

The margin, though, has a third component that is explicitly temporary. Raspberry Pi built strategic inventory ahead of the memory price rises. That inventory was bought at the old cost and, for part of the half, sold into boards priced at the new one. The result is a gross margin that flatters the underlying manufacturing economics. Management did not hide this. The guidance language attached to the full-year upgrade says adjusted EBITDA is expected ahead of market consensus despite moderation in the exceptional unit economics seen earlier in the year — which is a company telling you, in a results release, that the best part of the margin is behind it.

The demand signal alongside it is more durable. Growth was led by OEM and industrial customers rather than the hobbyist channel, which is the segment that carries multi-year design wins and recurring volume. The 2.6m-unit backlog is the visible expression of that: orders placed against a supply chain that management described as challenging, not against a promotional cycle.

Why the market reacted the way it did

A 16% single-day move on a company that had already roughly doubled in 2026 is not a response to a beat alone. It is a response to a guidance change with a specific shape. Coming into the print, the bear case on Raspberry Pi was straightforward: a hardware business with thin structural margins, exposed to a memory cost shock it could not fully pass on, trading on a multiple that assumed it could. The results answered the middle clause. Raspberry Pi did pass the cost on, it passed it on with enough pricing power to expand margin rather than merely defend it, and it did so while still shipping 17% more units. That combination — price increases that do not destroy volume — is the single most valuable piece of evidence a component business can produce.

The full-year upgrade then converted a half-year datapoint into a forward number. Investors were not being asked to extrapolate; they were being handed a company-sanctioned figure for the whole year. That is a different quality of information, and it is why the move was large.

The qualifier in the guidance wording is doing more work than the size of the move suggests it was given credit for. The exceptional unit economics have already moderated: the company says the lower-cost inventory has been largely consumed and that current inventory costs now sit closer to market prices, which will weigh on gross profit per unit in the second half. So the second half carries lower incremental margins on similar or higher revenue. The upgrade is still an upgrade, but the path to it runs through a narrowing spread between what Raspberry Pi pays for memory and what it charges for a board.

The Openbook read

Momentum is unambiguously the strongest of the five factors and is the one most likely to be over-read. A more-than-doubling of the share price this year and a 16% single-session gain will score at or near the top of the range on any trend-following construction. That is a description of what has happened, not a forecast, and momentum scores of this magnitude are the most mean-reverting input in the framework. Treat it as the factor carrying the least information about the next twelve months.

Growth improves materially and deservedly. Ninety per cent revenue growth, 17% unit growth and a 2.6m-unit backlog are real. The nuance is composition: a growth score that reads 90% as organic volume expansion is scoring a price pass-through as demand. The defensible growth number here is closer to the unit line plus the mix shift into silicon and compute modules — still good, and still accelerating, but a long way below the headline. A framework that cannot separate those two will overstate this factor for as long as memory pricing stays elevated.

Profitability is where the work is. Adjusted EBITDA margin on the half is roughly 16% of revenue, against about 14% a year earlier, and profit before tax nearly quadrupled. Both are improvements. But part of the improvement is the inventory timing effect, and management has said the effect is fading. The question for the score is whether the structural margin — boards manufactured at current memory prices, sold at current board prices — sits above or below where it was before the memory move. On the disclosure available, that cannot be resolved from the outside, and any profitability score treating the H1 margin as the new baseline is making an assumption the company has declined to make.

Solvency is the quiet factor and the one worth watching for a different reason. A strategic inventory build is working capital, and working capital is cash. Buying memory ahead of a price rise was the right commercial call and it is visible in the margin; it also means cash conversion in the half will look different from profit. A business growing units 17% into a constrained supply chain, while carrying a 2.6m-unit backlog, has a structural need to hold stock. That is a balance sheet characteristic, not a one-off, and it belongs in the solvency read rather than being netted off against a strong EBITDA line.

Reward/risk is the factor the other four feed into, and after a run that has more than doubled the shares this year it is the least comfortable. The reward case is a company with pricing power, OEM design wins, a visible backlog and an upgraded full-year outlook. The risk case is that the multiple now embeds a margin the company has flagged as moderating, in a business whose largest input cost is set by a commodity market it does not control. Those are not symmetric in timing: the demand evidence arrives over years of design cycles, while the memory-cost evidence arrives every quarter. Investors can screen the full five-factor picture on the Openbook screener.

The read-across

The obvious peer read is upstream. If Raspberry Pi's average selling price rose 42% largely because memory got more expensive, the memory makers are on the other side of that trade. Micron and its peers have been the beneficiaries of the same DRAM cycle that is compressing Raspberry Pi's structural margin. Anyone holding both is, to a degree, hedged; anyone reading Raspberry Pi's margin as evidence of a general hardware upcycle has the causation backwards.

The second read is across to the wider embedded and edge-compute complex. Raspberry Pi's industrial and OEM growth is a datapoint on demand for low-cost local compute, and the company's own framing has leaned on edge AI. That is relevant to Arm and to the broader set of names selling into industrial automation and connected devices, because it suggests the volume is being pulled by deployment rather than by evaluation kits. It is one company's half-year, so it is a datapoint rather than a trend, but it points the same way as the backlog.

For the UK market specifically, Raspberry Pi is now one of the few London-listed technology businesses producing this kind of growth rate, and its index weight has moved accordingly. That creates its own dynamic: a stock that has doubled in a market short of technology exposure attracts flow for reasons unconnected to memory pricing.

What to watch next

• Full-year results. The upgrade is for the year to December 2026, so the full-year print is the first hard test of whether the moderating unit economics arrive as a gentle taper or a step down. The gap between H1 adjusted EBITDA of USD 40.3m and the full-year outcome is the cleanest available measure of how much of the H1 margin was timing.
• The second-half gross margin line specifically. Revenue can be read off the backlog; margin cannot. Watch whether ASP holds near USD 65.90 now that the cheaply bought inventory is largely consumed, and what that does to the spread.
• Unit growth versus the 2.6m backlog. A backlog converts or it ages. Shipments running ahead of 17% growth in the second half would confirm the OEM demand story independently of price; shipments flat against a stable backlog would suggest supply, not demand, is the binding constraint.
• DRAM spot and contract pricing. This is the exogenous variable and it is observable between reporting dates. A continued rise supports ASP but squeezes input cost; a reversal does the opposite on both sides, with the timing determined by inventory turns.
• Mix disclosure. Any further breakout of silicon and compute module revenue against boards is the disclosure that would let the market separate durable growth from pass-through, and it would materially change how the growth and profitability factors should be scored.

Raspberry Pi has produced the strongest results in its public history and has been unusually straight about which part of them was a windfall. The interesting work from here is not celebrating the 90%; it is watching what the margin looks like when the windfall is gone.