What’s funding like in the AI chip market right now?

All funding deals in the AI chip market over the last 24 months

Below is the table listing all the deals. You can find our methodology at the end of this page.

If you want a deeper understanding of the market and its current dynamics, get our report covering the AI Chip Market.

Is AI chip funding actually healthier today than a year ago?

AI chip funding is much healthier today for companies that are ready to scale, while the early end of the market has clearly become tougher.

We counted 29 financings involving 21 companies in the latest 12-month period, compared with 11 financings across 10 companies one year earlier. Disclosed capital reached $9.88 billion, up from $1.82 billion, and the median round jumped from $44.6 million to $250 million.

The catch is where that improvement sits. Capital raised through rounds of $50 million or less fell 39%, and early-stage deals dropped from almost two thirds of the market to less than one quarter.

More companies can attract serious money now, and the market can support many more large financings at once. The improvement is much harder to see among smaller and earlier rounds.

Metric	Latest 12 months	Previous 12 months	Change
Financings	29	11	+164%
Funded companies	21	10	+110%
Disclosed capital	$9.88B	$1.82B	+443%
Median round	$250M	$44.6M	+461%
Capital from rounds ≤$50M	$63.5M	$104.5M	−39%
Early-stage deal share	24.1%	63.6%	−39.5 pp
Later-stage deal share	65.5%	27.3%	+38.2 pp
Zero-deal months	1	4	−3

Is the AI chip funding boom still real without the monster rounds?

AI chip funding still looks stronger after removing the very biggest deals, but most of the dollar boom disappears once we move below $100 million.

The headline increase is huge: disclosed capital rose 443%. Yet funding from rounds of $100 million or less increased only 8%. Go below $50 million and the comparison turns negative.

Removing the largest deal does not kill the story either. The biggest current financing represents 11.1% of the period's capital, versus 41.2% previously. The top three fell from 86.2% of capital to 31.4%.

In practical terms, the market now supports a whole group of giant rounds rather than depending on one or two exceptional deals. Cerebras raised $1.1 billion and then another $1 billion. SambaNova completed a $1 billion first close. Etched raised $300 million and then $700 million within weeks. Positron announced an $875 million financing structure.

Large checks are doing almost all of the work, but those checks are spread across far more companies.

Outlier test	Latest 12 months	Previous 12 months
Total disclosed capital	$9.88B	$1.82B
Capital excluding >$100M rounds	$271.7M	$251.1M
Capital excluding >$50M rounds	$63.5M	$104.5M
Largest deal / total capital	11.1%	41.2%
Top 3 / total capital	31.4%	86.2%
Top 5 / total capital	47.3%	94.3%
Deals above $100M	22	3
Share of deals above $100M	75.9%	27.3%

Are more AI chip companies actually getting funded now?

Yes, AI chip funding is reaching many more companies now, and activity has also become much more consistent through the year.

The number of funded companies more than doubled, from 10 to 21. We also found activity across five AI chip categories instead of three and across four regions instead of three.

The monthly pattern changed as well. The previous period had four rolling months with no qualifying financing. The latest period had only one. Average deal activity increased from 0.92 to 2.42 financings a month.

That is a pretty big change in breadth. Several different groups are finding capital at the same time: established players such as Cerebras and SambaNova, fast-scaling startups such as Etched and Positron, European entrants such as Fractile and OLIX, and Asian companies including Rebellions, FuriosaAI and Sunrise.

The broader company base is real. The harder question is what kind of company can access that money.

Are new AI chip startups still getting a shot?

New AI chip startups are still getting funded, although most of today's capital is flowing after companies have already survived their first financing.

We found four first financings in the latest period versus one in the previous year. The median first round rose moderately, from $26 million to $33.3 million.

The total for first financings looks much more dramatic at $549.5 million, but Unconventional AI's $475 million seed round explains most of it. That deal is an extreme case and tells us very little about the normal entry price for an AI chip startup.

Follow-on financing tells a different story. Companies that had already raised before attracted $9.33 billion, and their median round reached $275 million.

New companies can still break into AI chips. The really deep pools of money currently open up later, once investors have something more concrete to evaluate.

New entrants vs existing companies	Latest 12 months	Previous 12 months
First financings	4	1
First-financing share	13.8%	9.1%
First-financing capital	$549.5M	$26M
Median first financing	$33.3M	$26M
Follow-on financings	25	10
Follow-on capital	$9.33B	$1.79B
Median follow-on financing	$275M	$51.6M

Are investors mostly doubling down on AI chip winners?

AI chip investors are doubling down much more aggressively on companies that have already cleared the first technical and commercial hurdles.

Eight companies raised more than once during the latest 12 months. Only one did so in the previous period.

Cerebras returned after its $1.1 billion round for another $1 billion. Rebellions followed its $250 million Series C with a $400 million pre-IPO financing six months later. Positron moved from a $230 million Series B into a far larger Series C structure. Etched managed the most extreme sequence, announcing $300 million in July and another $700 million in August.

The interesting part is that these repeat rounds are spread across several companies. Investors are clearly concentrating dollars on proven names, but the group of companies considered worthy of that treatment has widened.

There are simply more companies capable of coming back to the market for another very large check than there were a year ago.

Have $100 million AI chip rounds become normal now?

Yes. A $100 million AI chip round has moved from exceptional territory into the mainstream of the current funding market.

Twenty-two of the 29 latest financings exceeded $100 million. One year earlier, only three of 11 did.

The middle of the distribution has shifted with them. As seen above, the median financing now sits at $250 million, which means giant rounds are shaping the typical deal rather than merely pulling up the average.

At the same time, the old middle has almost vanished. Financings between $20 million and $50 million represented more than a quarter of previous deals and only one current transaction.

That shift fits the economics of the companies raising today. Building a semiconductor architecture is expensive. Moving into advanced packaging, manufacturing, memory supply, servers, racks, software, data centers and customer deployments pushes the capital requirement much higher.

Has AI chip funding moved away from Seed and Series A?

AI chip funding has moved sharply away from Seed and Series A and toward companies already entering serious commercialization.

Seed financings currently represent 13.8% of deals and Series A another 10.3%. Together, they accounted for 63.6% of the previous market.

Series B, Series C, Series D+ and growth rounds now make up most of the activity. Even Series B has become unusually large, with a $230 million current median.

MatX's $500 million Series B, Fractile's $220 million Series B and OLIX's $312 million Series B show how far hardware financing can stretch before a company reaches a conventionally "late" stage. At the other end, Series D+ rounds currently have an $850 million median.

Stage names therefore tell only part of the story. The amount of capital attached to each stage has moved upward too.

Stage	Latest deal share	Previous deal share	Latest median round
Seed	13.8%	27.3%	$33.3M
Series A	10.3%	36.4%	$110M
Series B	17.2%	0%	$230M
Series C	17.2%	9.1%	$275M
Series D+	20.7%	18.2%	$850M
Growth Equity	10.3%	0%	$147M
Unknown	10.3%	9.1%	$169M

Are AI chip investors taking more risk, or just backing more proven companies?

AI chip investors are writing much bigger checks now, but those dollars are generally going to more proven companies rather than to a wave of speculative newcomers.

The median funded company is four years old today, compared with two years in the previous period. Among follow-on recipients, median age increased from 2.5 to four years.

We still see some extraordinary early bets. Unconventional AI raised $475 million at Seed, while several young companies have secured nine-figure Series A or B rounds. AI hardware can force investors to put a lot of money behind companies early simply because tapeouts and systems development are expensive.

Across the whole market, though, capital has moved toward companies with longer histories, previous investors and more evidence that the technology can leave the development phase.

Investors appear willing to take huge financial risk once some of the technical and commercial uncertainty has been cleared away.

Do AI chip startups need customers before they can raise big now?

AI chip startups can still raise before broad commercial adoption, but the biggest rounds increasingly arrive after investors have seen working silicon, real deployments or serious customer commitments.

Etched is a good example. Its $300 million financing was presented as capital for production and customer deployments. Weeks later, the company announced another $700 million alongside its first customer rack at Jane Street and more than $1 billion in customer contracts.

Positron's latest financing followed a deployment of more than 50 Atlas racks at Oracle Cloud Infrastructure. The money is now supporting its Asimov tapeout and Titan production ramp.

FuriosaAI offers a longer progression. RNGD moved from customer sampling to adoption by LG AI Research and then into volume production. Rebellions has similarly moved from chip development toward complete rack-scale products and commercial infrastructure deployments.

The bar varies by company, so we should not turn this into a rigid rule that every startup needs revenue before raising. What has clearly changed is the quality of proof around many of the largest financings. A promising architecture can get attention; a chip already moving toward real workloads can unlock a completely different amount of money.

Is inference really where AI chip funding is moving?

Inference is currently the clearest center of gravity in AI chip funding, mainly because existing inference companies have grown into much larger financing needs.

Inference accelerators already dominated the previous period, so the category did not suddenly take over by adding dozens of new startups. Its deal share actually stayed fairly stable, moving from 72.7% to 69.0%.

Round sizes changed much more. The median inference financing climbed from about $31.8 million to $240 million, with roughly $6.18 billion of current capital going into the category.

Etched is building frontier inference clusters. SambaNova is pushing its RDU infrastructure deeper into enterprise inference. Positron is scaling Atlas while developing its next silicon generation. FuriosaAI has moved RNGD into production. Rebellions now sells integrated inference infrastructure rather than only an accelerator chip.

Inference was already a popular technical thesis. Lately, it has been turning into a production and deployment story.

Is AI chip funding becoming less US-centric?

AI chip funding is becoming less US-centric, with Asia-Pacific taking a much larger share of activity and Europe attracting substantially bigger rounds.

North America still dominates dollars, accounting for 72.4% of current capital, but that is down from 89.1%. Its share of deals also slipped from 54.5% to 48.3%.

Asia-Pacific moved in the opposite direction. Its deal share rose to 37.9% from 27.3%, while its capital share increased to 19.9% from 8.9%.

Several different companies are behind that change. South Korea has Rebellions and FuriosaAI, while China contributed financings for Sunrise, AIGCIC and other domestic semiconductor companies. Public and strategic capital are also becoming more visible around parts of the Asian ecosystem.

Europe looks different. Its deal share fell, yet the money per deal became much larger. Fractile raised $220 million, while OLIX raised $220 million and later $312 million.

North America remains the financing center for now, but the rest of the market is no longer a rounding error.

Are strategic investors playing a much bigger role in AI chip funding?

Strategic investors are much more involved in AI chip funding today, especially around companies moving into manufacturing, infrastructure and customer deployment.

We identified strategic participation in 18 of the 29 latest financings, compared with four of 11 previously. That moves the participation rate from 36.4% to 62.1%.

The names make the trend easier to understand. Arm invested around companies including OLIX and Rebellions. SK Hynix backed Etched as the company prepared to scale production. Intel Capital appears around SambaNova. M12 participated in inference-hardware rounds. Jane Street went further by becoming both an Etched investor and its first disclosed rack customer.

Rebellions' $250 million Series C is another useful example because Arm, Samsung Ventures, Pegatron VC and Jusung Engineering connect the financing directly to processor IP, semiconductor manufacturing, electronics and production equipment.

Our investor records do not capture every participant in every round, so the totals below should be read as documented lead and strategic participation rather than a complete census of the investor universe. Even with that limitation, the change is large.

Investor measure	Latest 12 months	Previous 12 months
Deals with strategic participation	18	4
Strategic participation rate	62.1%	36.4%
Median strategic-backed financing	≈$306M	≈$409M
Distinct named lead/strategic organizations	≈61	≈22
Named organizations appearing on multiple deals	5	0
Lead investor disclosed	75.9%	81.8%
Inference deals among latest strategic-backed rounds	13 of 18	—

Are the same AI chip companies raising again unusually fast?

Some AI chip companies are raising again extremely quickly now, although the market-wide fundraising cycle has only shortened modestly.

Among current follow-ons where we could identify an earlier financing in the same market dataset, the median gap is 7.2 months. The previous-period median was 7.9 months.

That difference by itself is too small for a sweeping claim. The previous comparison also rests on only three usable observations, so we would be giving it more precision than it deserves.

The eye-catching part is the cluster of very fast repeat raises. Etched returned within weeks. Cerebras, SambaNova and Sunrise had sequences around four months, while Rebellions came back after roughly six.

When customer demand or production requirements change quickly, investors currently seem willing to reopen the financing conversation almost immediately.

Are follow-on AI chip rounds getting much bigger than the previous round?

Yes. AI chip companies that successfully return to investors are often raising multiples of their previous financing.

Across 16 current follow-ons with a usable earlier round size, the median step-up is 2.68 times. About 69% of those financings are more than twice as large as the previous known round.

The company trajectories make that easier to grasp. Positron moved from a roughly $52 million Series A to $230 million at Series B before announcing its much larger Series C structure. Fractile went from a $15 million seed round to $220 million at Series B. Rebellions followed $250 million with $400 million six months later.

The older comparison gives us too few observations to say confidently that step-ups across the entire market have accelerated by a specific amount.

What we can say with much more confidence is that successful AI chip companies are climbing the funding ladder in very large jumps these days. Their next problem often costs several times more to solve than the previous one.

What are AI chip companies spending these huge rounds on now?

AI chip companies are increasingly spending new funding on production, complete systems, deployments and commercial expansion alongside continued chip R&D.

The change is visible in how companies themselves describe their plans. Cerebras said its $1.1 billion financing would support processor, packaging and system development while expanding US manufacturing and data-center capacity. SambaNova's current business is built around deploying complete RDU-based inference infrastructure rather than simply developing a semiconductor.

OLIX says it is moving toward its first racks. Fractile is building both chips and systems for frontier-model inference. FuriosaAI has been financing manufacturing and go-to-market expansion while continuing work on its next-generation chip.

The most recent company updates fit the same lifecycle. FuriosaAI entered volume production and has been expanding commercial operations across Asia-Pacific. Rebellions has moved deeper into rack-scale infrastructure and international deployment.

R&D still consumes enormous amounts of capital in this industry. What has changed is everything that comes after R&D. Once a chip company starts manufacturing at volume, integrating servers, securing memory and packaging, supporting customers and expanding into new countries, the funding requirement can jump by hundreds of millions of dollars.

Is AI chip funding becoming a more mature market?

Yes, AI chip funding currently looks much more like a maturing infrastructure market than the early venture market we were looking at a year earlier.

Company age is moving upward. Later stages now account for most financings. Repeat rounds are common. Strategic investors are appearing more often. The biggest checks increasingly sit next to manufacturing, rack-scale systems, customers and production deployments.

Recent operating developments reinforce that reading. FuriosaAI moved RNGD into volume production and is expanding its regional commercial footprint. Rebellions has been extending its infrastructure into additional markets and customer environments. Etched has gone from working silicon to a rack running at Jane Street. Positron has already placed dozens of racks inside Oracle Cloud Infrastructure.

There is plenty of technical risk left, and several companies are still years away from proving durable economics. But fundraising is increasingly rewarding progress further down the industrial curve.

The AI chip market has grown up quickly.

What's the most misleading AI chip funding headline right now?

The most misleading AI chip funding headline would be that a fivefold increase in capital means fundraising became easier for every AI chip startup.

That headline would capture the spectacular part of the market and miss the difficult part.

Cerebras, SambaNova, Etched, Positron and several peers show how much money is currently available when a company reaches the right stage. At the same time, funding below the large-round threshold has not kept pace, and early-stage activity has lost a large share of the market.

As pointed out above, even the rise in company count does not mean capital has become equally accessible at every point in the startup lifecycle. Investors are funding more companies while setting aside most dollars for a relatively mature group.

For founders, the current market can therefore feel exceptionally strong or surprisingly narrow depending on where the company sits. Once a business reaches production, deployments or convincing commercial validation, the financing ceiling is extraordinarily high. Before that point, the picture is much less generous.

What really changed over the last 12 months?

The biggest change is that AI chip funding developed a real scale-financing layer. Nine-figure rounds now appear across a broad group of companies rather than around one isolated winner, and several businesses can absorb hundreds of millions of dollars in quick succession.

Breadth improved too. More companies raised money, more regions were active and the year contained far fewer quiet months. Today's market has much more depth than the previous one.

The improvement stops being broad once we look at smaller rounds. Early funding lost share, and capital below the large-growth-round threshold remained weak. The current boom therefore favors companies that have already crossed important technical or commercial hurdles.

Inference has benefited most clearly from this maturation. Investors were already interested in inference a year ago; what changed is that companies such as Etched, Positron, Rebellions, FuriosaAI and SambaNova are now financing manufacturing, systems and deployments rather than only chip development.

Strategic capital has become part of that transition. Semiconductor companies, memory suppliers, infrastructure groups and customers increasingly sit next to financial investors, which makes sense as the bottleneck shifts toward getting hardware manufactured and deployed at scale.

The funding bar has moved with the market. Working silicon still matters, but the largest rounds increasingly come with stronger evidence around customers, production, complete systems or real workloads.

The clearest description of AI chip funding today is a market that has become much bigger, broader and more mature at the top while remaining difficult at the bottom. Looking only at the $9.88 billion headline would miss the most important part of the change.