Arteris
Arteris sells the designs for the network that carries data around inside other companies' chips, and earns a royalty on every chip shipped.
The scores
Growth3/4 ↑
Revenue up 46.2% in Q2 2026 and 22.3% in 2025, after two years near 7%; part of 2026 was bought with Cycuity.
Business quality3/4 →
Customers return design after design and NXP and AMD keep buying, but in-house teams and Arm's bundle do the same job.
Financial position4/4 ↑
$123.3M of cash and investments, no debt, and $6.8M of free cash flow over the year, so no further share sale is needed.
Execution record3/4 →
Beat the 2025 revenue guide and raised 2026 twice, but landed at the worst end of the 2025 loss guide and widened 2026's.
Per share capture2/4 ↓
Shares up 11% in seven months on a share sale and a share-paid deal, with stock pay a quarter of revenue.
Potential2/4 ↑
The slice it sells into was about $109M in 2019 inside a market growing 12.7%, and in-house teams hold much of the rest.
Risk profile3/4 →
China is 24.5% of revenue, some of it from Entity List parties; no customer is above 10%.
Valuation2/4 ↑
At 12.9 times sales the price needs 26% yearly growth and a margin swing to 30% after stock pay by 2029.
Evidence quality3/4 →
Revenue, cash and shares are filed, but the market size is from 2019, Cycuity's revenue is undisclosed and the big wins are unnamed.
The Verdict
We rate Arteris neutral. It designs the network that lets the parts of a chip talk to each other, licenses it to chipmakers and earns a royalty on each chip shipped; customers rarely switch, so the business is solid. But it still loses money once the shares it pays staff are counted, and its share count keeps rising. The price needs about a quarter more sales every year and a healthy profit it has never earned, so it is full, not cheap. The view hinges on losses ending while share pay shrinks and growth stays fast. The growth is real; the profit is still a promise.
The long caseChips keep gaining more parts that must talk to each other, so chipmakers buy that network rather than build it; Arteris must turn that demand into profit.
The evidence under this - 9 claims
The vendor's market value of $1.14B uses about 46.2M shares, before the at-the-market sale, so vendor multiples read about 6% low (S08, S03).
- Revenue grew 22.3% in 2025 and is guided up 37% in 2026, so 26% a year is below the current pace (C04, C07).
- Arteris has never earned an operating profit, and stock pay alone is a quarter of revenue, so the margin half of the requirement has no precedent in its own record (C20, C21).
- The adjusted loss guide for 2026 widened twice while revenue was raised, so costs ran faster than the model in the same year it was restated (S02, S06, S04).
- The first half had cost growth close to revenue growth rather than half of it, 38.7% against 42.5% (S03).
The company
A modern chip holds processors, memory and specialised engines that constantly pass data to each other. Arteris sells the design for the network that carries that data: it packs each request into small packets and routes them over shared wires.
Chipmakers license the design rather than build it themselves. NXP calls Arteris's products integral to chips it is developing, and AMD uses it inside its AI chips beside its own connection technology. Getting this wrong is costly: only 14% of chip projects work first time.
The evidence under this - 6 claims
How it makes money
A customer pays a licence fee per chip design, usually over two to three years, and Arteris engineers help put the network into silicon. Once the chip ships, one to five years later, a royalty arrives on every unit, for seven years or more.
Licences and support were $20.8M of Q2 2026's $24.1M revenue. Royalties are still only about a tenth of revenue, but they are growing faster, and $131.4M of signed work had not yet been counted as revenue.
The evidence under this - 4 claims
Where the money comes from, stream by stream
| Stream | Share of sales (H1 2026) | How it earns it |
|---|---|---|
| Licensing, support and maintenance | 85.2% | Two-to-three-year licences to use its on-chip networks and chip-assembly software in a set number of chip designs, with support and engineering help. |
| Variable royalties | 9.8% | A fee per chip, or a share of its price, once a customer ships chips containing Arteris interconnect. |
| Professional services and other | 5.0% | Mostly Cycuity's hardware-security work for the U.S. government, done by subcontractors at a much lower margin. |
Who runs it
Charles Janac has run Arteris since 2005 and chairs its board; a new finance chief, Saurabh Sinha, started in September 2026. Insiders own about 23%, almost all of it Janac's. His yearly bonus pays on goals the company does not name, and his share awards vest just for staying, with no profit or share-price test.
The evidence under this - 4 claims
Janac and Bayview Legacy each sold 100,000 shares under 10b5-1 plans on 1 September 2026, and insiders made no purchases in the year (C27, S11).
Why it wins, and who could beat it
Why customers come back
Once a chip family is built around Arteris's network, its settings, testing and, in cars, safety certification are tied to it. Switching means redoing that work on the next design, so customers tend to come back. Design starts rose from 76 to 83 in 2025.
The price is small against the risk: a complex chip design costs about $12.2M, and a faulty network can mean starting again. One crack: Mobileye, a large past customer, now holds rights to another company's network patents.
The evidence under this - 7 claims
- Retention of about 90% is stated without a definition or a period, and the largest 2025 customer fell from 15% of revenue to below 10% in the first half of 2026 (S01, S03).
- The cost is paid per new chip project, not per chip shipped, so a customer can move at the next design without touching chips already in production (S03).
The growth forecast dates from 2017 and Semico's current pages could not be read, so the rate may have moved with AI chips (S16).
The threat
The biggest rival is the customer's own engineers. Qualcomm and Intel each bought a network business to bring the job in-house. Arm sells its own network and bundles one with its processor designs, Cadence launched one in 2024, and Baya, a funded start-up, already shares a customer with Arteris.
The evidence under this - 6 claims
- Arm already ships interconnect as part of its Compute Subsystems, so the largest third-party rival is inside the processor licence buyers already sign (S09).
- The entry-time estimate is management's own and is not measured by anyone outside the company (S01).
The rivals: QCOM, INTC, ARM, CDNS, MBLY, and 2 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| QCOM · INTC | Customers' own interconnect teamsPriced against it: Different product | Interconnect IP (network-on-chip licences and royalties) | On sale | Different product |
| ARM | Arm CoreLink NI-700 and the interconnect inside Arm's Compute SubsystemsPriced against it: Small part of it | Interconnect IP (network-on-chip licences and royalties) | On sale | Small part of it |
| CDNS | Cadence Janus network-on-chipPriced against it: Different product | Interconnect IP (network-on-chip licences and royalties) | On sale | Different product |
| Private | Baya SystemsPriced against it: Private | Interconnect IP (network-on-chip licences and royalties) | On sale | Private |
| MBLY · INTC | Intel's network-on-chip patents, sublicensed to MobileyePriced against it: Different product | Interconnect IP (network-on-chip licences and royalties) | On sale | Different product |
| In-house | Customers' own security verification and general formal toolsPriced against it: In-house | Software: SoC integration automation and Cycuity security verification | On sale | In-house |
How big it could get
How big this could get
Nobody publishes the size of this market today. The last outside count, for 2019, put it near $109M, with Arteris holding about a third.
The wider market for chip designs sold as licences was $9.45B in 2025, and the part holding these networks grew 22.2% that year, faster than the whole. The number of complex chip projects grows slowly, so the growth comes from each chip needing more.
The evidence under this - 4 claims
- The figure is seven years old and reaches us only through Arteris's own release, so it carries the seller's framing (S15).
- It counts merchant licences only; the work in-house teams do, Arteris's main rival, is not in it (S01).
The growth forecast dates from 2017 and Semico's current pages could not be read, so the rate may have moved with AI chips (S16).
Who has signed
$131.4M of contracted work cannot be cancelled, $67.5M of it due within a year. AMD has licensed the newest network for its AI chips. Management says a large cloud company and a chip design house signed big deals in Q2 2026, but names neither and gives no size.
The evidence under this - 4 claims
- Neither buyer is named, the size of either deal is not disclosed, and no customer reached 10% of revenue in the first half (S03, S05).
- The chief executive said the design-house buyer was previously a relatively small customer, so the standardisation is new and unproven in revenue (S05).
Where the growth is coming from, business by business
Where the growth is coming from
| Business | Share of revenue | Growth: last quarter | Last half-year | Last year | Share in 3 years at its half-year rate |
|---|---|---|---|---|---|
| License and Maintenance | 87% | +25.1% | +21.5% | +20.9% | 77% |
| Royalty | 13% | +75.1% | +53.9% | +49.7% | 23% |
| Service, Other | 0% | +127.3% | +30.2% | -75.4% | 0% |
| Whole company | 100% | +30.0% | +24.3% | +22.3% | 100% |
On the filed split through 2025-12-31, if each business keeps its last half-year's growth, the company grows about 26% next year and 28% in year 3, against 33.8% over the last twelve months. A scenario, not a forecast: no business holds its rate.
How we read it: The split is by revenue type, not business: royalties grow faster but are a tenth of revenue. The score reads the filed total, which includes Cycuity, so it believes neither blend fully.
The results
The numbers
Sales grew 34% over the twelve months to Jun 2026, faster than the 22% in the year to Dec 2025.
Operating margin is −48%, up from −65% in the year to Dec 2023.
Running costs take 135¢ of each sales dollar, down from 148¢ in the year to Dec 2021.
Shareholders' money lost 131% a year, down from −114% in the year to Dec 2021.
It turned 8.0% of sales into free cash, $6.77M over the last twelve months.
Yearly sales growth has ranged from +6.5% to +33% over the last 5 years.
The share count rose 115% in 4.5 years, and pay in shares costs 25% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-04. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Jun 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $37.9M | $50.4M | $53.7M | $57.7M | $70.6M | $84.6M |
| Sales growth | +19% | +33% | +6.5% | +7.6% | +22% | +34% |
| Profit | ||||||
| Gross margin | 90% | 92% | 91% | 90% | 90% | 88% |
| Operating margin | −57% | −57% | −65% | −55% | −47% | −48% |
| Net margin | −62% | −54% | −69% | −58% | −49% | −47% |
| What growth costs, per sales dollar | ||||||
| Research and development | 81% | 82% | 84% | 78% | 71% | 67% |
| Sales and marketing | 31% | 35% | 38% | 36% | 38% | 38% |
| General and admin | 35% | 32% | 33% | 30% | 29% | 27% |
| Pay in shares | 15% | 23% | 27% | 28% | 26% | 25% |
| Capital spending | 2.1% | 2.1% | 2.8% | 0.6% | 2.0% | 2.1% |
| Cash | ||||||
| Cash from operations | −$814K | −$6.77M | −$15.7M | −$720K | $6.73M | $8.51M |
| Free cash flow | −$1.62M | −$7.82M | −$17.2M | −$1.04M | $5.34M | $6.77M |
| Balance sheet | ||||||
| Cash and short-term investments | $85.8M | $68.2M | $41.2M | $43.8M | $33.9M | $93.3M |
| Returns and shares | ||||||
| Shares (diluted) | 22.0M | 32.6M | 35.7M | 38.9M | 42.3M | 47.3M |
What growth costs
Growth is slowly making Arteris less unprofitable: research and office costs rise more slowly than sales, so each new dollar of sales loses less. Selling costs and pay in shares still grow as fast as sales, and the company still loses money on every dollar it sells.
The evidence under this - 2 claims
The last report (6 Aug)
Quarter ended June 2026, filed figures.
Revenue rose 46% to $24.1M, and the yearly value of contracts plus royalties reached $99.5M, up 44%. Management raised its 2026 revenue guide for the second time, to $95M to $98M.
Part of that was bought. Cycuity, acquired in January, does government security work through subcontractors, which pushed the share of revenue kept after direct costs from 89% to 85%. The expected 2026 loss before stock pay also widened, to $7M to $10M, even as revenue was raised.
What it did to the thesisThe print confirmed demand is growing and the recurring base is rising, but it left the central question open: costs grew almost as fast as revenue, so whether a profit ever arrives after stock pay is no clearer than before.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $-0.10 against $-0.04 expected; $-0.30 as filed, the gap being what the company adds back.
The evidence under this - 6 claims
- Management says much of the August raise came from Cycuity's government work, which carries a much lower gross margin, so the raise is partly bought revenue (S05).
- The 2025 adjusted loss landed at $12.5M, the bad end of an $8.5M to $12.5M guide, so the profit half of the guide has slipped before (S07, S02).
- It is an unaudited company metric that now includes Cycuity's contracts, so part of the 44% was bought (S03, S06).
- Recognised revenue over the last four quarters was $84.6M, well short of the $99.5M run rate, because fixed fees are spread over multi-year terms (S08).
- The adjusted loss guide for 2026 widened twice while revenue was raised, so costs ran faster than the model in the same year it was restated (S02, S06, S04).
- The first half had cost growth close to revenue growth rather than half of it, 38.7% against 42.5% (S03).
What management did with the money
Arteris sold 2.1M new shares for $75.0M and paid for Cycuity partly in shares. It now holds $123.3M of cash and no debt, but each existing share owns a smaller slice of the company.
Insiders in 2026Q3: 10 purchases and 89 sales, from their Form 4 filings.
The evidence under this - 4 claims
The price, and where we disagree
In short. Full. The price pays for fast growth and a jump to healthy profits that Arteris has never earned, with little room if either falls short.
The shares more than doubled over the year, then fell 57.6% from their 30 June high to 14 September with no bad news from the company. The business kept growing through the fall.
A new finance chief started on 8 September, and the Q3 2026 results on 3 November are the first test of the raised guide and of management's hope for a profit before stock pay as soon as Q4.
The evidence under this - 4 claims
- Management says much of the August raise came from Cycuity's government work, which carries a much lower gross margin, so the raise is partly bought revenue (S05).
- The 2025 adjusted loss landed at $12.5M, the bad end of an $8.5M to $12.5M guide, so the profit half of the guide has slipped before (S07, S02).
How we read the multiple
Buyers pay about 12.9 times the last year's sales for a company that has never made an operating profit. Working back:
- Company value of about $1.09B, grown 10% a year, is about $1.45B in 2029.
- At 25 times profit, a mature licensor's price, that needs about $58M of 2029 operating profit after share pay.
- At a 30% margin, that is about $194M of sales: 26% growth a year.
- Today that margin is minus 57.6%.
Against its own history. Well above where the shares usually traded, because a year of excitement about AI chips raised what buyers pay for each dollar of sales.
The evidence under this - 4 claims
- Revenue grew 22.3% in 2025 and is guided up 37% in 2026, so 26% a year is below the current pace (C04, C07).
- Arteris has never earned an operating profit, and stock pay alone is a quarter of revenue, so the margin half of the requirement has no precedent in its own record (C20, C21).
The vendor's market value of $1.14B uses about 46.2M shares, before the at-the-market sale, so vendor multiples read about 6% low (S08, S03).
Priced against its rivals
None of the listed rivals above sells the same thing to the same buyers as a large part of what it does. Arm Holdings does too, but as a small part of a much bigger company, so its price says little about this one. A price comparison needs at least two, so the price is judged against its own history instead.
Also checked and set aside: Synopsys (a substitute, not the same product).
Private or foreign rivals that cannot be priced: Baya Systems, Customers' in-house interconnect teams.
Where we disagree with the market
A growing AI chip winner whose losses are about to end, as six buy ratings and one hold say.
The growth is real, but the losses come from paying staff in shares, and that cost is not shrinking.
Analysts look at the loss before share pay, which management expects to turn into a profit as soon as Q4 2026. Counting share pay, the Q2 2026 loss was 57.6% of revenue, and share pay alone was 26% of 2025 revenue.
We would be wrong if a profit before share pay arrives in Q4 while share pay falls below a fifth of revenue and growth stays above 25% once Cycuity is in both years.
What analysts say, and the evidence
The evidence under this - 5 claims
- Management says much of the August raise came from Cycuity's government work, which carries a much lower gross margin, so the raise is partly bought revenue (S05).
- The 2025 adjusted loss landed at $12.5M, the bad end of an $8.5M to $12.5M guide, so the profit half of the guide has slipped before (S07, S02).
- The adjusted loss guide for 2026 widened twice while revenue was raised, so costs ran faster than the model in the same year it was restated (S02, S06, S04).
- The first half had cost growth close to revenue growth rather than half of it, 38.7% against 42.5% (S03).
What could go wrong, and how we would know
The risks
- Paid in shares. Share pay was 26% of 2025 revenue, and shares outstanding rose 11% in seven months.
- Customers can build it. The biggest chipmakers keep their own network teams, and Qualcomm and Intel each bought one to stop paying.
- China is a quarter. China was 24.5% of 2025 revenue and barely grew, and some sales go to companies on a U.S. export blacklist.
- A big customer may leave. Mobileye now holds rights to another company's network patents and no longer names Arteris in its annual report.
- Costs keep pace. Costs rose almost as fast as revenue in early 2026, and the expected loss was widened twice this year.
The evidence under this - 7 claims
- Arm already ships interconnect as part of its Compute Subsystems, so the largest third-party rival is inside the processor licence buyers already sign (S09).
- The entry-time estimate is management's own and is not measured by anyone outside the company (S01).
- The adjusted loss guide for 2026 widened twice while revenue was raised, so costs ran faster than the model in the same year it was restated (S02, S06, S04).
- The first half had cost growth close to revenue growth rather than half of it, 38.7% against 42.5% (S03).
What we still do not know
- How much of 2026 revenue Cycuity brought. Arteris calls the deal too small to report separately.
- Whether Mobileye's newest driving chips still use Arteris's network.
- How big this market is today. The research firm that counts it sells the figure.
- Which cloud company standardised on Arteris, and how large the deal is.
The evidence under this - 4 claims
- The figure is seven years old and reaches us only through Arteris's own release, so it carries the seller's framing (S15).
- It counts merchant licences only; the work in-house teams do, Arteris's main rival, is not in it (S01).
- Neither buyer is named, the size of either deal is not disclosed, and no customer reached 10% of revenue in the first half (S03, S05).
- The chief executive said the design-house buyer was previously a relatively small customer, so the standardisation is new and unproven in revenue (S05).
What would change our view
- Neutral→AttractiveA price that survives a margin shortfall, with the run rate still rising
It needs all of:
- ○The shares fall until the company is valued at 9 times sales, leaving room for profits to fall short.about ≈$17.20 today, 30% below the last close
- ○The yearly value of contracts plus royalties reaches $106M or more by the end of 2026.checked at the next results, due in Nov 2026
- Neutral→UnattractiveGrowth slows, or the price outruns it
It needs any of:
- ○Arteris guides 2027 revenue growth below 15%.checked at the next results, due in Nov 2026
- ○The shares rise until the company is valued at 18 times sales, which would need about 41% growth a year.about ≈$34.41 today, 40% above the last close
We would also be wrong if
- Fewer chip designs start on Arteris in 2026 than the 83 of 2025.
- NXP, AMD or another top customer names a different network supplier, or its own.
- Royalties grow less than 20% a year for two quarters in a row.
What is coming
- 3 November 2026 Q3 2026 results: Revenue against the $24M to $25M guide, the adjusted loss against $3M, and whether the raised 2026 guide holds under the new finance chief.
Every dated event ahead
- 3 November 2026 Q3 2026 results: Revenue against the $24M to $25M guide, the adjusted loss against $3M, and whether the raised 2026 guide holds under the new finance chief.
- February 2027 Q4 and full-year 2026 results and the 2027 guide: Whether Q4 brings the first adjusted operating profit management floated, what the Cycuity earn-out pays on 2026 bookings, and whether 2027 costs grow at half the pace of revenue.
Since we wrote this
- 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 3 OctThesis writtenNeutral, medium convictionWritten two months after the Q2 2026 print, after a 57.6% fall from the June high and a partial rebound.
Also checked 3 times with nothing new to act on.
Sources 30
Sources - 30
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-02-12Item 1 products, model, competition and switching costs; revenue 2024-2025; one customer at 13%; geography with China 24.5%; Entity List disclosure; stock pay $18.4M; 170 engineers; 83 design starts; options and RSUs outstandingbacksC01C02C03C04C12C13C14C15C16C20C21C22C23C26C50C54
- S02primarycompany release2026-02-12the first 2026 guide ($89M to $93M revenue, $5M to $9M adjusted operating loss), royalties $6.6M up 50%, four billion chips shipped, FlexGen at 10 customersbacksC07C09C10C12C36
- S03primarycompany filing2026-08-06revenue by type, Cycuity purchase terms, the at-the-market sale, balance sheet, share count at July 30, RPO of $131.4M with $67.5M due in twelve months, customer concentration, design starts, the reasons cost of revenue doubledbacksC01C03C05C06C08C12C15C21C22C23C25C26C33C49C50C54
- S04primarycompany release2026-08-06ACV plus royalties $99.5M, RPO $135M, adjusted results, the raised 2026 guide and the Q3 guide, named winsbacksC05C07C08C09C10C18C20C34C35
- S05primarytranscript2026-08-06Q2 2026 earnings call transcript (FMP)data/cache/transcripts/AIP-2026Q2.json (FMP earning-call-transcript, quarter label 2026Q2, call held 2026-08-06)management's account of the hyperscaler and design-house wins, the French payroll tax, the royalty growth target, the Q4 profit hope; questions from five analysts, a separate origin from the answers
1 of the world's largest hyperscale cloud companies has chosen to adopt and
backsC11C18C19C22C24 - S06primarycompany release2026-05-12the second 2026 guide ($91M to $95M revenue, $4.5M to $8.5M adjusted operating loss), ACV plus royalties $92.8M, a hyperscaler and a memory maker as winsbacksC07C19
- S07primarycompany release2025-02-18the 2025 guide: revenue $66M to $70M, adjusted operating loss $8.5M to $12.5M, free cash flow $1M to $7MbacksC36
- S08secondaryvendor data2026-10-03Research pack (FMP market tier, FINRA short interest, daily closes, SEC XBRL statements)the 2 October close, 52-week range, price-move split, quarterly series, analyst split and its drift, short interest settled 15 SeptemberbacksC04C20C28C31C32C33C34
- S09primarycompany filing2026-05-26Arm Holdings Form 20-F, fiscal year ended March 31, 2026 Arm Holdings plcArm's own description of its Compute Subsystems, which include interconnect, and of interconnect as part of its platformbacksC17C41
- S10primarycompany filing2026-02-19Cadence Design Systems Form 10-K, fiscal year 2025 Cadence Design Systems, Inc.the IP portfolio Cadence describes to investors; the words network-on-chip and NoC appear nowhere in itbacksC40
- S11primarycompany filing2026-10-03Arteris Form 4 filings, October 2025 to September 2026, read by insider_fetch.py210 open-market sales and no purchases; the Rule 10b5-1 flag on eachbacksC27C35
- S12primarycompany filing2025-12-11the $75M at-the-market programme announced the day before the 17.2% fallbacksC33C54
- S13primaryindustry data2026-09-04the size and growth of the design IP market in 2025, System IP growth of 22.2%, and the 74.4% held by Arm, Synopsys and Cadence; no dollar figure for interconnect
the worldwide design IP market grew 12.7% in 2025 to reach $9,451.9 million.
backsC37 - S14primaryindustry data2025-04-14the 2024 base of the design IP market
Design IP revenues achieved $8.5B in 2024 and this is an all-time-high growth of 20%.
backsC37 - S15secondarycompany release2020-06-09the only public figure for the interconnect slice, through Arteris's framing: 31.3% on $34M in 2019, second to Arm
established a 31.3% share of the interconnect IP market as a result of its $34 million in 2019 revenues.
backsC38 - S16primaryindustry data2017-04-20the slow growth of complex SoC design starts, from a measurer; old
Semico is forecasting that the Compound Annual Growth Rate (CAGR) for designs of Advanced Performance Multicore SoCs will be in the 3.2% range over the next 5 years.
backsC39 - S17primarycompany release2024-06-26Cadence release, 'Cadence Janus Network-on-Chip IP' (Design & Reuse) Cadence Design Systems, Inc.Cadence's entry with a first-generation NoC, coherence on its roadmap
The Cadence Janus NoC will be available in July 2024.
backsC40 - S18primaryother2026-10-03Arm product page, CoreLink NI-700 network-on-chip interconnect Arm Holdings plcArm's own non-coherent NoC and its automotive safety version, NI-710AE
CoreLink NI-700 A scalable and highly configurable packetized Network-on-Chip (NoC) with a 30 percent wire count reduction compared to a traditional crossbar.
backsC41 - S19secondarynews2026the assets in the ARC sale, none of them an interconnect
the assets moving across include ARC-V (RISC-V), ARC “Classic” CPU IP, VPX DSP, and NPX NPU lines, plus associated software and ASIP tooling (ASIP Designer and ASIP Programmer).
backsC48 - S20primarycompany release2025-01-23Baya Systems Series B release (silicon.co.uk) Baya Systems (private)a funded NoC start-up backed by Synopsys and Intel Capital, and its first named customer
Early Baya customers and partners include Tenstorrent, which has licensed Baya technology for its AI and RISC-V chiplet solutions
backsC42 - S21secondarynews2013-11-06Qualcomm taking the FlexNoC technology and two-thirds of the staff in-house
Qualcomm acquired the technology and the engineering team, but is letting Arteris continue licensing the technology to anyone, including Qualcomm’s competition.
backsC43 - S22secondarynews2018-09-10a second chipmaker buying a merchant NoC vendor
Intel acquired NetSpeed Systems, taking in network-on-a-chip and interconnect fabric intellectual property
backsC43 - S23primarycompany filing2026-02-12Mobileye Global Form 10-K, fiscal year 2025 Mobileye Global Inc.a network-on-chip patent sublicence through Intel in June 2024; Arteris not namedbacksC44
- S24primaryindustry data2025-09-03what a failed chip integration costs the industry, measured across projects
we found that only 14% of ASIC/SoC projects achieved first-silicon success - the lowest figure in more than twenty year
backsC45 - S25secondarynews2026-02-15NXP's own executive on its use of FlexNoC, Ncore and CodaCache
Arteris is a trusted technology provider whose silicon-proven products are integral to our advanced solutions under development today.
backsC46 - S26secondarynews2025-08AMD's own vice-president on FlexGen inside its chiplets
Integrating Arteris’ FlexGen NoC IP technology into a range of AMD chiplets, we can automate interconnect configuration and enable seamless connectivity among SoC components
backsC47 - S27primarycompany filing2025-12-22Synopsys Form 10-K, fiscal year ended October 31, 2025 Synopsys, Inc.the IP Synopsys sells, which names interfaces, memory, security and processors and no network-on-chipbacksC48
- S28secondarynews2024-11-06Tenstorrent as an Arteris licensee, from Arteris's release
network-on-chip (NoC) IP for on-chip connectivity in its chiplet-based products.
backsC42 - S29primarycompany filing2025-11-04AMD's additional licence orders after its multi-project FlexGen licence, Altera's expansion and 2V SystemsbacksC47
- S30primarycompany filing2026-04-21directors' and officers' biographies and since-dates; the 2025 bonus design and payout; the 2025 equity grants and their vesting; the beneficial ownership table at 9 April 2026backsC51C52C53