What are the top semiconductor startups by total funding raised?

Top startups in the Semiconductor Market ranked by total funding raised

Here is an updated table that ranks the top startups in the Semiconductor Market based on the total amount of funding they have raised to date.

The table also includes the total number of funding rounds, the date and size of the latest round, the financing type (e.g. Series A, equity financing), key investors, the startup’s current status (active, IPO, acquired, or shut down), and a confidence score based on the data collected.

If you need to dig deeper and get more detailed data, please check our report covering the Semiconductor Market.

Which semiconductor startups have raised the most funding?

ChangXin Memory Technologies, Yangtze Memory Technologies and Cerebras currently sit far above the rest of the semiconductor funding market, with roughly $19.1 billion, $10.2 billion and $9.3 billion raised respectively in our reconstructed dataset.

The gap is huge. CXMT alone has raised more than the combined total of companies ranked fourth through eighth. The top three have attracted about $38.6 billion between them, or roughly 42% of the disclosed capital across the 95 companies we reviewed.

Their funding stories are very different, though. CXMT and YMTC manufacture memory chips in China, where fabs and production expansion require billions of dollars. Cerebras developed wafer-scale AI processors and then raised more than $6 billion from its public listing after a $1 billion private round earlier in the year.

That difference matters throughout the ranking. A $10 billion manufacturing company and a $1 billion fabless chip designer may appear in the same table, but the money is paying for very different businesses.

Rank	Company	Main market	Total raised
1	ChangXin Memory Technologies	DRAM memory	$19.1B
2	Yangtze Memory Technologies	NAND memory	$10.2B
3	Cerebras Systems	AI compute	$9.3B
4	Horizon Robotics	Automotive AI chips	$3.9B+
5	Hua Hong Semiconductor	Semiconductor foundry	$3.3B
6	GTA Semiconductor	Semiconductor foundry	$3.1B
7	Groq	AI inference infrastructure	$2.8B
8	Nexchip	Semiconductor foundry	$2.7B
9	UNISOC	Mobile and IoT chips	$2.1B
10	Moore Threads	GPUs	$1.7B+

Why has ChangXin Memory Technologies raised so much money?

ChangXin Memory Technologies has raised so much because CXMT is building DRAM manufacturing capacity on a scale that looks more like heavy industry than a normal semiconductor startup.

The biggest jump came from its Shanghai IPO. CXMT raised about RMB57.9 billion, or roughly $8.6 billion, before any overallotment. Reuters described it as Asia's largest IPO of the year at the time and the largest Chinese A-share semiconductor offering ever.

That single transaction was larger than the cumulative funding of almost every company in our semiconductor ranking.

CXMT also had years of private and state-linked financing before listing. Reuters' review of company filings found that investors linked to the Hefei government collectively held 36.8% of the company around the IPO. The company originally grew out of a Hefei-backed effort to establish domestic DRAM production.

The scale makes more sense once we remember what CXMT has to finance. DRAM production requires fabs, manufacturing equipment, process development, engineering teams and repeated capacity upgrades. A fabless AI chip startup can outsource manufacturing; a memory manufacturer cannot outsource the core reason it exists.

So CXMT's $19.1 billion cumulative total is better read as evidence of what it costs to build a domestic memory champion than as evidence that venture investors suddenly started writing $10 billion checks to chip startups.

Is Cerebras now one of the most-funded semiconductor startups ever?

Yes. Cerebras is now one of the most heavily funded semiconductor challengers we can identify, although its giant cumulative total now includes more than $6 billion raised through its IPO.

Cerebras entered the year with a substantial private financing history, then raised another $1 billion in a Series H at a post-money valuation of roughly $23 billion. Tiger Global led that round, with Benchmark, Fidelity, AMD, Coatue, Atreides and several other investors participating.

A few months later, Cerebras sold 34.5 million shares at $185 in its IPO after the underwriters exercised their full option. The company reported approximately $6.38 billion in gross proceeds.

Those two transactions alone added about $7.4 billion.

Cerebras has also become a much more infrastructure-heavy business than its original description as a "wafer-scale chip startup" suggests. Its systems are sold for AI training and inference, and the company has been building much larger deployment relationships around its Wafer Scale Engine.

The headline funding figure is real, but the reason it jumped is simple: Cerebras moved from venture-backed semiconductor company to a publicly financed AI infrastructure business.

How concentrated is semiconductor startup funding?

Semiconductor funding is extremely concentrated: the three largest companies in our dataset account for about 42% of the roughly $92.7 billion of disclosed capital we reconstructed.

The concentration stays high further down the ranking. The top five absorb about half of the visible funding. The top ten represent almost 63%. By the time we reach the 20 largest companies, they account for roughly three quarters of the total.

The median tells the story even better. Average funding across the 95 companies comes to roughly $976 million, while the median is only about $289 million.

A handful of giant transactions pull the market average far above what a typical company in the dataset actually raised.

Companies included	Approx. funding	Share of total
Largest company	$19.1B	20.6%
Top 3	$38.6B	41.6%
Top 5	$45.8B	49.4%
Top 10	$58.2B	62.8%
Top 20	$70.7B	76.3%
Top 50	$86.2B	92.9%
All 95	$92.7B+	100%

Are the most-funded semiconductor startups actually still private startups?

No. Most of the companies in this ranking have already gone public, been acquired or shut down, so the table is really a funding history of semiconductor challengers rather than a list composed only of private startups today.

Among the 95 companies, 39 are public and 20 have been acquired. One, GreenWaves Technologies, is classified as shut down. That leaves 35 active private companies.

Public companies also dominate the capital total. They account for around $58.7 billion of the disclosed funding in our dataset, or roughly 63%. Active private companies represent around $29.5 billion.

That explains why names such as CXMT, Cerebras, Hua Hong, Horizon Robotics and Moore Threads sit so high. Their totals include money raised after reaching public markets.

Status today	Companies	Approx. disclosed funding
Public	39	$58.7B
Active private	35	$29.5B
Acquired	20	$4.6B
Shut down	1	$28M
Total	95	$92.7B+

Which private semiconductor startups have raised the most money?

Among companies still classified as private and active, GTA Semiconductor, Groq and UNISOC lead our dataset, followed by Positron AI, Ayar Labs, Tenstorrent and Rebellions.

GTA Semiconductor sits around $3.1 billion. Groq is around $2.8 billion on our cumulative reconstruction, although Groq's corporate structure and strategy changed substantially after its licensing agreement with Nvidia. UNISOC follows at approximately $2.1 billion.

The newer generation gets more interesting just below them.

Positron has now disclosed more than $1.15 billion from three major rounds since 2025: more than $50 million in its Series A, $230 million in its Series B and $875 million in its Series C.

Ayar Labs has moved above $1 billion on the company's own latest disclosures. Its $500 million Series E took total funding to $870 million, and the company then added another $150 million.

Rebellions says its $400 million pre-IPO round brought cumulative funding to $850 million. Tenstorrent has raised more than $900 million based on our reconstructed history.

These private totals show where investors can still place very large semiconductor bets before an IPO.

Company	Main focus	Approx. cumulative funding
GTA Semiconductor	Specialty foundry	$3.1B
Groq	AI inference infrastructure	$2.8B
UNISOC	Mobile and IoT chips	$2.1B
Positron AI	AI inference processors	$1.15B+
Ayar Labs	Optical chiplets	$1.02B+
Tenstorrent	AI accelerators / RISC-V	$927M
Rebellions	AI inference	$850M+
Lightmatter	Photonic interconnect	$850M
SemiDrive	Automotive chips	$608M
MatX	AI training accelerators	$605M

Are AI chip startups raising more money than traditional semiconductor startups?

Yes, the biggest new private semiconductor checks are currently going disproportionately to AI compute, inference, networking and photonics.

Positron's latest round was $875 million. MatX raised $500 million. Rebellions raised $400 million. SiFive raised $400 million as it pushed RISC-V further into data-center computing. Groq raised $650 million and then another $350 million after its restructuring. d-Matrix raised $275 million. Axelera AI raised more than $250 million.

Traditional chip categories still attract serious money. SemiDrive and Calterah continue to raise capital for automotive processors and radar. Foundries and memory manufacturers remain capable of raising far more than AI startups once public markets and state financing are included.

The split is sharpest in private growth capital. Investors are currently much more willing to put $250 million to $875 million into a company tied to AI infrastructure than into a conventional wireless, MCU or consumer semiconductor startup.

What are the biggest recent semiconductor startup funding rounds?

The largest recent private semiconductor rounds are approaching the size of small IPOs, with Positron's $875 million Series C standing out among the companies we reviewed.

Positron raised $875 million at a $5 billion valuation after a $230 million Series B earlier in the year. That gave the company more than $1.1 billion of newly disclosed financing in roughly seven months.

MatX raised $500 million in a Series B led by Jane Street and Situational Awareness. Ayar Labs raised $500 million in its Series E and later secured another $150 million. Rebellions closed a $400 million pre-IPO round. SiFive also raised $400 million.

Groq's financing deserves separate treatment because the company changed direction after licensing technology to Nvidia. It raised $650 million in growth capital and then another $350 million in a newly labelled Series A.

A step below those mega-rounds, d-Matrix closed $275 million and Axelera AI more than $250 million.

Company	Recent financing	Amount
Positron AI	Series C	$875M
Groq	Growth capital	$650M
MatX	Series B	$500M
Ayar Labs	Series E	$500M
Rebellions	Pre-IPO	$400M
SiFive	Growth round	$400M
Groq	Series A after restructuring	$350M
d-Matrix	Series C	$275M
Axelera AI	Growth financing	$250M+

Why is Positron AI raising so much money so quickly?

Positron AI is raising unusually large amounts because investors are betting that AI inference needs purpose-built hardware and that Positron can reach production before that market settles around a few incumbents.

The speed is remarkable. Positron disclosed a $50 million-plus Series A in 2025, a $230 million Series B earlier this year and an $875 million Series C more recently. Its own funding history therefore shows more than $1.15 billion raised across those three rounds.

The latest round valued Positron at $5 billion.

The money is funding a difficult jump from systems built around first-generation hardware into its next custom silicon. Positron says more than 50 Atlas racks have already been deployed at Oracle, while its next-generation Asimov processor is supposed to move the company toward much larger inference deployments.

Positron's pitch also targets one of the biggest cost problems in current AI systems: memory. The company is trying to build inference hardware around large memory capacity and lower-cost memory architectures instead of depending entirely on the same HBM-heavy approach used by leading GPUs.

That strategy still has to prove itself at large commercial scale. Investors, though, have already shown they are willing to spend more than $1 billion before waiting for the final answer.

Is Groq still mainly an AI chip startup?

Groq today looks increasingly like an AI inference infrastructure company built around a valuable chip architecture, rather than a straightforward independent challenger trying to beat Nvidia processor for processor.

The turning point came when Groq entered into a non-exclusive technology licensing agreement with Nvidia. Founder Jonathan Ross and other senior people moved to Nvidia, while Groq continued operating separately.

Groq then raised $650 million to expand its AI inference cloud. The company said it was already operating 13 data centers across several regions and serving more than five million developers at that point.

It followed that with another $350 million Series A for the restructured business, valuing Groq at $3.5 billion. The company said the two rounds together represented $1 billion in recent capital.

That makes Groq harder to compare directly with an early-stage company still raising money mainly to develop and commercialize a new processor.

Why are photonics, networking and power semiconductor startups raising so much money now?

Photonics, networking and power semiconductor startups are raising much more money because AI systems are running into physical bottlenecks around data movement and power delivery, not just raw compute.

Ayar Labs provides the clearest recent example. The company raised $500 million to expand high-volume production and testing for its co-packaged optics technology. That round brought total funding to $870 million. Ayar later added another $150 million, taking cumulative financing above $1 billion.

Lightmatter had already shown how valuable investors considered the same bottleneck. Its $400 million Series D valued the company at $4.4 billion and brought total funding to around $850 million.

Then the acquisition market confirmed the strategic value. Marvell completed its purchase of Celestial AI, another optical-interconnect company, for total purchase consideration of roughly $3.5 billion.

Other companies attack adjacent constraints. Retym develops coherent DSPs for AI networks. Enfabrica focused on networking silicon. Empower Semiconductor developed high-density power delivery technology and was later acquired by Analog Devices for $1.5 billion. Navitas and Innoscience are working on advanced power semiconductors.

The common issue is straightforward: larger AI clusters need processors to exchange more data while consuming less energy per bit and handling much higher power density. Capital is following those bottlenecks.

Has Ayar Labs now raised more money than Lightmatter?

Yes. Based on the latest company disclosures, Ayar Labs has moved slightly above Lightmatter in cumulative funding, with more than $1 billion raised versus roughly $850 million for Lightmatter.

The change happened quickly.

Ayar's $500 million Series E brought its cumulative financing to $870 million. It then secured another $150 million, which takes the company to roughly $1.02 billion based on its disclosed totals.

Lightmatter's website currently states about $850 million of total funding. Its biggest round remains the $400 million Series D that valued the company at $4.4 billion.

The investor mix is revealing. Ayar's recent financing included AMD, Nvidia, MediaTek and Alchip alongside large institutional investors and QIA. Those companies sit directly inside the semiconductor and AI infrastructure ecosystem Ayar wants to supply.

Lightmatter has attracted investors such as Fidelity, T. Rowe Price and GV, while also building manufacturing relationships with GlobalFoundries, TSMC, Tower Semiconductor, Amkor and ASE.

The fundraising race between Ayar and Lightmatter has moved well beyond proving that silicon photonics works. Both now need to prove they can manufacture and integrate optical technology at the volumes required by future AI systems.

Are automotive semiconductor startups still attracting big funding rounds?

Automotive semiconductor startups are still raising meaningful capital, but the really huge private rounds have shifted toward AI data-center infrastructure.

Horizon Robotics remains the standout automotive company in the overall ranking with more than $3.9 billion of reconstructed financing. It has also been able to tap public markets after years of private funding.

SemiDrive remains private at around $608 million in our dataset and recently added a $100 million Series C. Calterah, which develops automotive radar chips, has raised more than $150 million and completed another substantial round.

Black Sesame Technologies has raised more than $660 million and is now publicly traded. Arbe Robotics and Uhnder have attracted roughly $240 million and $195 million respectively around automotive radar.

The comparison with AI is stark. A $100 million automotive semiconductor round currently looks substantial; Positron raised almost nine times that amount in one transaction.

Who is investing the most in semiconductor startups right now?

Semiconductor funding now comes from a mix of venture firms, institutional investors, sovereign funds and chip companies, and the biggest rounds increasingly combine several of those groups.

Traditional venture capital is still everywhere. NEA appears around Positron and Mythic. Spark Capital backed companies including MatX, Enfabrica and Retym. Kleiner Perkins has invested in Alif Semiconductor and Retym.

Large asset managers have become especially important once semiconductor companies reach later stages. Fidelity appears repeatedly across the dataset. T. Rowe Price led Lightmatter's $400 million round. BlackRock has participated in several large hardware financings.

Strategic chip companies are also becoming much more visible. Nvidia has participated in Ayar Labs and SiFive and plans to participate in Groq's latest financing. AMD invested in Ayar and Cerebras. MediaTek appears in Ayar and led HyperLight's recent round. Samsung-linked capital has backed several AI and edge semiconductor companies.

Sovereign investors have moved deeper into the same deals. QIA participated in Positron, Ayar Labs, d-Matrix and HyperLight, covering AI accelerators, photonic chiplets and other infrastructure technologies. Chinese state-linked funds are even more important in memory, foundries and domestic processor companies.

Nvidia deserves separate attention because its influence now goes beyond minority investing. The Groq relationship became a technology-licensing and talent deal, showing how a strategic investment relationship can become something much deeper.

Investor group	Examples	Where the money is going
Venture capital	NEA, Spark, Kleiner Perkins	AI chips, networking, edge silicon
Asset managers	Fidelity, T. Rowe Price, BlackRock	Large late-stage rounds
Chip companies	Nvidia, AMD, MediaTek, Intel	Photonics, RISC-V, AI infrastructure
Sovereign investors	QIA, GIC	AI chips and infrastructure
Chinese state capital	National and regional semiconductor funds	Memory, foundries, GPUs, domestic chips
Corporate venture funds	Samsung Ventures, Qualcomm Ventures, GV	Connectivity, processors, photonics

Are governments and sovereign funds changing semiconductor startup funding?

Yes. Government-linked capital has become a major part of semiconductor financing because chips are now treated as strategic infrastructure in China, the United States and several other countries.

China provides the clearest example. The country's third major national semiconductor investment fund was established with RMB344 billion of registered capital. Regional governments and state investment vehicles also appear repeatedly across companies in our ranking.

CXMT shows how powerful that model can become. Reuters found that Hefei government-linked investors collectively controlled 36.8% of the memory company around its IPO.

The U.S. uses a different structure. CHIPS Act support mostly reaches semiconductor companies through manufacturing incentives rather than normal startup equity rounds. The U.S. Government Accountability Office reported recently that Commerce had funded 49 projects across 24 companies and disbursed $13.1 billion by April.

Sovereign wealth funds are joining private rounds directly. QIA has recently invested in Positron, Ayar Labs, d-Matrix and HyperLight, covering AI accelerators, photonic chiplets and other infrastructure technologies.

That makes semiconductor funding harder to read than software funding. A large round can reflect venture conviction, industrial strategy, national-security priorities or all three at once.

How much semiconductor funding now comes from IPOs?

IPOs and other public equity financings account for a huge part of the capital at the top of our ranking, which is why the largest totals look much bigger than ordinary venture funding statistics.

CXMT raised roughly $8.6 billion in its Shanghai IPO before overallotments. Cerebras generated about $6.38 billion in gross proceeds. Hua Hong's A-share IPO raised roughly $3 billion.

Moore Threads raised around $1.1 billion when it listed. ASR Microelectronics also raised roughly $1.1 billion in its IPO. MetaX, GigaDevice, Iluvatar CoreX and several other companies have used public markets to add hundreds of millions more.

Companies can also return to public investors later. Horizon Robotics completed a large post-IPO placement. Montage Technology, Navitas Semiconductor, Espressif Systems and Arbe Robotics have all raised additional equity after becoming public.

For semiconductor companies, cumulative external equity can therefore keep rising long after the private-startup phase ends.

Does raising more money make a semiconductor startup more likely to win?

No. Semiconductor history is full of heavily funded companies that built impressive technology without becoming large independent winners.

Graphcore is one of the clearest examples. The company raised roughly $682 million in our reconstruction and at one point reached a multibillion-dollar private valuation. SoftBank eventually acquired it in 2024.

Nuvia raised around $293 million before Qualcomm bought it for approximately $1.4 billion. Rivos raised roughly $376 million before being acquired. Enfabrica raised more than $260 million before its acquisition.

Some outcomes were much stronger financially. Celestial AI had raised roughly $589 million in our dataset before Marvell bought it for about $3.5 billion. Empower Semiconductor had raised about $215 million before Analog Devices agreed to pay $1.5 billion.

Funding tells us how much capital a company convinced investors to commit. The final outcome can still range from a valuable acquisition to a disappointing exit.

Which semiconductor startups had the strongest acquisitions compared with the money they raised?

Celestial AI, Empower Semiconductor and Nuvia stand out as companies where strategic buyers ultimately paid several times the amount of capital the startup had raised.

Celestial AI raised roughly $589 million in our dataset. Marvell later reported about $3.5 billion of total purchase consideration for the acquisition. Celestial's photonic interconnect technology gave Marvell a direct route into scale-up connectivity for large AI systems.

Empower Semiconductor raised approximately $215 million before Analog Devices agreed to acquire it for $1.5 billion in cash. The logic was specific: AI chips are becoming harder to power efficiently as compute density rises, and Empower developed technology for delivering power closer to the processor.

Nuvia raised about $293 million before Qualcomm acquired it for roughly $1.4 billion. Qualcomm wanted Nuvia's custom CPU architecture and engineering team for future processors.

None of the three companies needed to become the next Nvidia to create a valuable semiconductor business. They built technologies that mattered enough to much larger buyers.

Are edge AI chip startups being left behind by the data-center funding boom?

Edge AI chip startups are still attracting serious funding, although the largest checks currently go to data-center AI companies.

Axelera AI recently raised more than $250 million. The company says it has attracted more than $450 million when equity, grants and venture debt are combined, although those categories should not all be counted as equity funding in a cumulative-equity ranking.

Alif Semiconductor has raised roughly $342 million in our dataset. SiMa.ai is around the same level. Hailo reached roughly $341 million before being acquired.

Kneron has raised around $190 million. EnCharge AI has raised about $144 million. Rain AI remains much smaller, around $41 million in our reconstruction.

These companies are trying to win markets such as industrial equipment, cameras, robotics and vehicles, where power consumption and cost per device matter enormously.

For now, those markets are producing smaller financing rounds than the AI data-center race.

Is China or the United States raising more semiconductor funding?

China dominates the very top of cumulative semiconductor funding, while the United States has a deeper cluster of newly funded private AI-chip and infrastructure startups.

China's numbers are enormous. CXMT has approximately $19.1 billion in our dataset, YMTC $10.2 billion, Horizon Robotics more than $3.9 billion, Hua Hong $3.3 billion, GTA Semiconductor $3.1 billion and Nexchip $2.7 billion.

Moore Threads, UNISOC, Biren, GigaDevice, Cambricon, MetaX and several other Chinese semiconductor companies add billions more.

That capital is spread across memory manufacturing, foundries, GPUs, automotive processors, mobile chips and other areas where China wants more domestic supply.

The United States has a different shape. Cerebras dominates the cumulative total after its IPO, while private capital is spread across Groq, Positron, Ayar Labs, Tenstorrent, Lightmatter, MatX, d-Matrix, Alif, SiMa.ai and many smaller architecture bets.

So China currently looks stronger on giant cumulative financing totals, especially once state-backed investment and IPOs are included. The U.S. private market is more fragmented and experimental, with investors financing several competing approaches to AI compute, inference and interconnect.

What does the semiconductor startup funding ranking actually tell us about the market today?

The semiconductor funding market today is being driven by two unusually powerful forces: China's push to build more of its own chip supply chain and the global race to build AI infrastructure.

The first force explains why CXMT, YMTC, Hua Hong, GTA and several other Chinese companies rank so highly. Memory fabs and foundries require billions of dollars, and public or state-linked capital can support projects far beyond normal VC scale.

The second explains the recent private rounds around Cerebras before its IPO, Positron, MatX, Groq, Rebellions, d-Matrix and Tenstorrent.

The money is also spreading beyond compute chips. Ayar Labs, Lightmatter, Celestial AI, Retym and Empower show how quickly optical connectivity, networking and power have become fundable categories of their own.

Our dataset also shows how misleading a simple average can be. Roughly $92.7 billion of disclosed capital sounds enormous, but the top three companies account for about 42% of it. The median company raised around $289 million, far below the nearly $1 billion average.

The clearest pattern is that semiconductor capital is abundant at the top but highly selective underneath. Investors, governments and strategic buyers are spending extraordinary amounts when a company sits directly on a major bottleneck such as domestic manufacturing, AI inference, memory or interconnect.