Mid-year refresh · July 2026

Top 10 Federal Grants for Canadian Tech Startups (2026)

Sixty federal programs in our catalog are open and tagged for technology companies; 34 of those are grants you never repay. The ten below are the ones a Canadian tech startup can realistically win this year, ranked by how winnable they are rather than by headline dollar size. NRC IRAP is the anchor — it funded 3,136 firms in FY 2024-25, and the median award in our catalog is about $75,000, not the $1 million ceiling. Mitacs Accelerate is the easiest first win. Almost everything here stacks with SR&ED.

Nine of the ten set no minimum-revenue bar, so pre-revenue founders qualify. The exception is CanExport SMEs, which requires $300,000 to $100 million in annual revenue and 3 to 500 employees — and whose 2026-27 window closes August 31, 2026.

Updated July 31, 2026 · Reviewed by Khalid Hamadeh, Founder · Every status, amount and date below re-checked against our grant catalog this week

What is actually dated right now

Most federal tech funding runs on rolling intake, which is why undated programs dominate this list. These four are the exceptions — the things on this page that expire.

Aug 31, 2026 CanExport SMEs closes its 2026-27 window (12:00 PM ET). You must allow 60 business days between submitting and your first funded activity, so anything booked before roughly late October is already out of scope.
Sep 15, 2026 Next SSHRC Partnership Engage competition (8:00 PM ET), then December 15. Your university partner submits, and their internal deadline lands 1–3 weeks earlier.
Oct 5, 2026 Full application for the NSERC Alliance Quantum SME Partner Stream. The letter of intent for this round closed July 27, 2026 — if you missed it, this cycle is shut and the next competition has not been posted.
No deadline IRAP, Mitacs Accelerate, Scale AI Acceleration, NSERC Alliance Advantage, NSERC ARD and the NRC AI for Productivity Challenge all take applications year-round. Rolling does not mean unlimited: IRAP regional budgets are fullest in April–May and can be committed by late fall.

How much do federal tech programs actually pay compared with their advertised maximums?

The median IRAP firm contribution is $75,000 against an advertised ceiling of $10 million; the median CanExport SME grant is $25,000 against a $50,000 cap; the median IP Assist contribution is $15,800 against $50,000. Across six federal programs with disclosed payments, the typical award sits between 0.75% and 50% of the maximum, and IRAP alone signed 2,507 contributions in 2025.

$75,000Median IRAP firm contributionMiddle half $50,000 to $194,900; $75,000 for a first award
$25,000Median CanExport SME grantHalf of the $50,000 cap; 2026 intake closed August 31
$15,800Median IP Assist contributionMiddle half $10,000 to $22,730
2,507IRAP contributions signed in 2025CanExport SMEs 1,674; IP Assist 101

Median award, middle half and advertised maximum, six federal programs (log scale)

Median awardMiddle half of awards (p25 to p75)Advertised maximum
Dots are medians, bands the middle half, and the marker the advertised maximum in the GrantCompass record. For-profit recipients only. CIIP mixes small partnership-development grants with large co-innovation projects, which is why its band is so wide. Hover or focus a row for the exact figures.

The gap between the ceiling and the median is the number to plan around. IRAP's $10 million maximum applies to large multi-year projects; three quarters of its contributions are under $194,900 and a first award has a median of $75,000. CanExport SMEs is the closest to its ceiling because it reimburses half of a capped expense list, so a $25,000 median simply means most applicants spend near the $50,000 eligible-cost maximum. FedDev Ontario's scale-up stream is the outlier in the other direction: a $1.75 million median, but as a repayable contribution to established firms, not a startup grant.

These figures are disclosed payments, not estimates, and they agree with the $75,000 IRAP median used throughout this guide. The sixty-program count above is this guide's technology-tagged federal shortlist; the six programs here are the ones among them whose payments are published agreement by agreement. SR&ED is a tax credit and does not appear in the disclosure, which is why it is absent from the chart rather than small.

Six federal programs: what they advertise and what they pay
Program (scope)Advertised maximumMedian awardMiddle halfAgreements in 2025
IRAP firm contributions$10,000,000$75,000$50,000 to $194,9002,507
CanExport SMEs$50,000$25,000$20,000 to $30,0001,674
IP Assist$50,000$15,800$10,000 to $22,730101
CanExport Innovation$37,500$11,777$6,090 to $19,838131
CIIP (co-innovation)$600,000$4,388$3,061 to $75,00097
FedDev Ontario BSP (repayable)$10,000,000$1,750,000$1,000,000 to $2,845,99713

Source: Government of Canada Proactive Disclosure of Grants and Contributions, for-profit recipient slices (IRAP firm contributions n = 17,641; CanExport SMEs n = 9,922; IP Assist n = 913; CanExport Innovation n = 596; CIIP n = 321; FedDev Ontario BSP n = 422), agreements to 2026; advertised maximums from the GrantCompass catalogue record for each program. Verified 2026-09-03. Contains information licensed under the Open Government Licence – Canada.

Who this list is for

If you're a pre-revenue AI founder

You have models, a demo, and a compute bill, but no meaningful revenue. Start with Mitacs Accelerate (#2) — a graduate researcher for $7,500 of your cash per four-month unit — then take Scale AI Acceleration (#3) through a certified partner accelerator for up to $50,000 of commercialization support. Once you have a defined R&D project with technical uncertainty, add NRC IRAP (#1) and claim SR&ED on the rest. Skip CanExport (#8) entirely: it requires $300,000 in annual revenue.

If you're a scaling SaaS company with real revenue

Past roughly $300,000 in annual revenue and 3+ employees, the list widens rather than narrows. CanExport SMEs (#8) becomes available for international expansion — and its 2026-27 window closes August 31, 2026. IRAP contributions scale with project size (the catalog's typical ceiling is $1 million, with a $10 million ceiling reserved for major capital projects). And SR&ED is where the largest single number usually sits: Budget 2025 raised the enhanced expenditure limit directly from $3 million to $6 million, putting up to $2.1 million a year in reach for a qualifying CCPC. Run the numbers in our SR&ED calculator before you chase a grant.

If you're a deep-tech or life-sciences startup with IP in development

You're building something technically ambitious — biotech, quantum, robotics, clean energy — with most team time in experimentation rather than selling. NSERC Alliance Advantage (#4) is the highest-leverage instrument on this page: roughly $2 of NSERC money for every $1 of your cash. IRAP (#1) covers your own payroll alongside it. Stack SR&ED on the spending neither one funded.

If you're exploring defence or dual-use technology

Autonomous systems, cybersecurity, sensors, advanced materials, secure comms. IDEaS Component 1 (#7) is the cleanest entry point in the country: up to $250,000 for the 1a concept stage with zero cost-share required, and a route into DND procurement that a grant alone doesn't buy. Challenges are posted 10 to 20 times a year through CanadaBuys, each with a 60 to 90 day window, so the work is watching for the right one.

If you've never applied for government funding before

Start with Mitacs Accelerate (#2) or SSHRC Partnership Engage (#5). Both carry the lowest application difficulty on this list and the highest approval rates in our catalog (both "High, above 40%"), and both build the organizational track record that reviewers of IRAP and NSERC Alliance look for later. Avoid Innovative Solutions Canada (#10) as a first application — our catalog flags it as not first-time-applicant friendly.

The ranked list of 10

Ranked by winnability, not by ceiling: application difficulty and approval rate first, then dollar size, then how much scaffolding (an academic partner, a consortium, a posted challenge) you need before you can apply at all. All ten were open to applications on July 31, 2026. Programs that are between intakes are listed separately at the bottom, not padded into the ten.

Tier 1 — Open, low barrier, no academic partner needed

Open · rolling Difficulty 3 / 5 Incorporation required

IRAP is Canada's most-used R&D grant for SMEs. An NRC Industrial Technology Advisor (ITA) is assigned to your company, assesses your project, and can approve contributions covering up to 80% of eligible R&D salaries and contractor costs. The number founders quote is the $1 million typical ceiling; the number that predicts your outcome is the median actual award of about $75,000, with a typical first-time SME landing between $75,000 and $200,000.

In FY 2024-25 IRAP reached 9,187 client firms and funded 3,136 of them. Intake is rolling, but the money is not evenly available across the year: regional budgets are fullest in April and May and can be committed by late fall in high-demand regions.

What to checkIRAP
Amount~$75K median · $75K–$200K typical first award · $1M typical ceiling
Application difficultyModerate (3/5)
Time to decision4 weeks (to $50K) to 9 weeks ($500K–$3M), after 2–4 months of ITA relationship-building
Stacks with SR&EDYes — the most common pairing in Canada

Verdict: The best single program for any incorporated Canadian tech startup doing genuine R&D is IRAP, because no other federal instrument combines rolling intake, up to 80% cost coverage, and an assigned advisor who tells you in advance whether your project will fly. Open the ITA conversation early in the fiscal year — the relationship is worth as much as the cheque.

Not for you if: you aren't incorporated, your work is commercial build-out with no technical uncertainty, or you've already started the R&D — IRAP cannot fund retroactively.

Source: National Research Council Canada — IRAP
Open · rolling Difficulty 2 / 5 Approval: high (>40%)

Mitacs Accelerate puts a graduate student or postdoc to work on your R&D problem. Each internship unit is worth $15,000 (standard graduate) or $20,000 (postdoctoral fellow), and you contribute half — about $7,500 in cash per standard unit. That's the whole trade: $7,500 buys roughly four months of a supervised researcher plus their professor's involvement.

Submit at least 8 weeks before the internship starts (16 weeks if international travel is involved). Decisions take 6 to 8 weeks. Provincial allocation restrictions that constrained Alberta and Ontario in FY 2024-25 were lifted in May 2025.

What to checkMitacs Accelerate
Amount$15K per unit (graduate); $20K (postdoc); ~$60K for a PhD over 4 units
Your share~$7,500 cash per standard unit (50%)
Application difficultyLow (2/5) — the lowest on this list
Stacks with SR&EDYes, on your own contribution

Verdict: The best first grant for a founder who has never applied for government funding is Mitacs Accelerate, because our catalog rates its difficulty 2 out of 5 and its approval rate "high, above 40%" — and a win here is the track record that IRAP and NSERC reviewers look for next.

Not for you if: you can't put up the cash share, you aren't incorporated, or the work has no research component a graduate supervisor would put their name on.

Source: Mitacs — Accelerate Program
Open · rolling Difficulty 2 / 5 AI / applied ML

Canada's AI supercluster funds up to $50,000 of acceleration support per startup, delivered through a network of certified partner accelerators and incubators. One detail decides whether this fits your cash-flow plan: you do not receive the money. The funding reimburses the partner accelerator for the eligible cost of supporting you — mentorship, IP work, commercialization services. Most participating startups receive the full $50,000 allocation in services.

Roughly 50 to 70 startups are supported per year across the partner network. Choosing the right partner matters more than the application: regional programs such as Volta in Halifax, Propel in Saint John or Amii in Edmonton often have less competition and more hands-on time than the largest urban accelerators.

What to checkScale AI Acceleration
AmountUp to $50,000 — paid to the accelerator, not to you
Application difficultyLow (2/5)
Time to acceptance2–8 weeks; 3–6 months to cohort start
Cost-shareNone

Verdict: The best low-effort program for an early AI startup is Scale AI Acceleration, because it is difficulty 2 out of 5, requires no matching cash, and buys commercialization help at exactly the stage when an AI team has a model and no go-to-market. Treat it as services, not runway.

Not for you if: you need cash in the bank, or your product has no applied-AI component — the supercluster's mandate is AI adoption and commercialization.

Source: Scale AI — Funding Programs

Tier 1 head-to-head: where the first $50,000 comes from

Swipe the table sideways to see every column.

IRAPMitacs AccelerateScale AI Acceleration
You receiveCash contributionA researcherAccelerator services
Your cash shareUp to 20% of eligible costs~$7,500 per unitNone
Effort to apply3 / 52 / 52 / 5

The take: these three are complements, not alternatives. A pre-revenue team with $10,000 to deploy gets the most out of Mitacs first, because it converts the smallest amount of cash into the most technical capacity — and the completed project becomes the R&D story IRAP wants to fund next.

Tier 2 — Open, but a university or college applies on your behalf

Open · rolling Difficulty 3 / 5 University applies

Alliance funds collaborative R&D between a company and a university researcher, at $20,000 to $1,000,000 per year for one to five years. The mechanic that makes it the highest-leverage instrument on this page is the ratio: every $1 of your cash unlocks roughly $2 from NSERC, so $50,000 of partner cash supports about $150,000 of total project funding, plus in-kind university time.

The structural fact most founders miss: your company does not submit the application. The university researcher does. Your job is finding the right faculty member, co-designing the research question, and committing the cash. Realistic first-time awards land at $20,000 to $300,000 per year. Small grants clear review in 5 to 9 weeks; larger ones take 9 to 24, plus up to 10 more if a national security assessment is triggered.

What to checkNSERC Alliance Advantage
Amount$20K–$1M/year for 1–5 years (realistic first award $20K–$300K/yr)
Leverage~2:1 — NSERC's $2 per $1 of your cash
Who appliesThe university researcher, not your company
Time to decision5–24 weeks depending on size

Verdict: The best option for a deep-tech startup that already knows a professor in its field is NSERC Alliance Advantage, because the 2:1 match is the most favourable cash multiplier in Canadian federal funding and a joint Alliance + Mitacs submission runs through a single coordinated review.

Not for you if: you have no academic relationship and no time to build one — sourcing the right researcher is realistically a one to three month exercise before anything is submitted.

Source: NSERC — Alliance Advantage Grants
Next: Sep 15, 2026 Difficulty 2 / 5 Approval: high (>40%)

Partnership Engage funds one-year social-science and humanities research done with a university partner on a problem you bring. The 2026 ceiling doubled from $25,000 to $50,000, which changes what's buildable: a $35,000 project now funds a researcher for six to twelve months of part-time work on your specific question. Between 200 and 320 awards are made a year across four cycles.

Deadlines are quarterly — March 16, June 15, September 15 and December 15, 2026, all at 8:00 PM ET — with results roughly 8 to 9 weeks later. Universities impose their own internal deadlines one to three weeks earlier. As with Alliance, the academic is the applicant of record.

What to checkSSHRC Partnership Engage
Amount$10K–$50K (ceiling doubled in 2026); $15K–$35K typical
Application difficultyLow (2/5)
Next deadlineSeptember 15, 2026, then December 15
Time to decision8–9 weeks

Verdict: The most underused program on this list for tech founders is SSHRC Partnership Engage, because user research, adoption studies, responsible-AI assessments and digital-transformation work all qualify as social-science research — and at difficulty 2 with an above-40% approval rate, it is the fastest federal money a startup can realistically win.

Not for you if: your question is purely technical. If there's no human, organizational or societal dimension, a faculty member won't be able to frame it for SSHRC — take it to NSERC instead.

Source: SSHRC — Partnership Engage Grants
Open · rolling Difficulty 3 / 5 College applies

ARD is Alliance's college-side counterpart: R&D partnerships with a college, CÉGEP or polytechnic instead of a university, funded up to $150,000 per year for three years. Colleges tend to be faster, more applied and more available to small companies than university labs. The business contributes about 25% cash and, as with Alliance, the institution submits the application, not you.

There is a real strategic seam in the review process: requests up to $40,000 per year are reviewed internally in 5 to 9 weeks, while anything from $40,001 to $150,000 goes to external peer review and takes 24 to 28 weeks. A first project sized just under the threshold gets you a funded partnership in two months instead of six.

What to checkNSERC ARD
AmountUp to $150K/yr for 3 years; $25K–$100K/yr typical
Your share~25% cash
Fast lane≤$40K/yr reviewed in 5–9 weeks (vs 24–28 above it)
Who appliesThe college applied-research office

Verdict: The best route for a hardware, manufacturing or applied-software startup with no university connection is NSERC ARD through a nearby college, because the applied-research office is set up to work with SMEs and the sub-$40,000 tier turns a funded partnership around in 5 to 9 weeks.

Not for you if: your work needs fundamental research capability — that belongs at a university under Alliance (#4).

Source: NSERC — Applied Research and Development Grants

Alliance vs ARD vs Partnership Engage: which academic partner

Swipe the table sideways to see every column.

Alliance AdvantageNSERC ARDSSHRC Engage
Partner typeUniversityCollege / polytechnicAny postsecondary
Ceiling$1M / year$150K / year$50K total
Fastest decision5 weeks5 weeks (≤$40K/yr)8–9 weeks

The take: all three make an institution the applicant, so the binding constraint is which institution will actually take your call. If you already have a professor, go Alliance for the ceiling. If you don't, a college applied-research office is materially easier to reach than a university lab — and ARD's sub-$40,000 tier is the fastest funded partnership in this group.

Tier 3 — Open, higher bar or a specific strategic angle

Open · challenge-based Difficulty 3 / 5 No cost-share

The Department of National Defence pays companies to solve defined defence and security problems. Component 1a pays up to $250,000 to design a solution concept; 1b pays up to $1,500,000 to develop it — and neither requires you to put up matching money. Successful projects can continue into Component 2, for a pipeline worth up to $6.75 million.

IDEaS issues 10 to 20 challenges a year, each with a 60 to 90 day application window, posted through CanadaBuys. The program explicitly invites civilian technology companies to reframe an existing product for a defence application, so the qualifying work is often positioning rather than new R&D.

What to checkIDEaS Component 1
AmountUp to $250K (1a) or $1.5M (1b); $150K–$250K typical for 1a
Cost-shareNone — 100% funded
Approval rateModerate (20–40%) per our catalog
Cadence10–20 challenges/year, 60–90 day windows, via CanadaBuys

Verdict: The best zero-dilution, zero-cost-share option for a startup with any dual-use angle is IDEaS Component 1a, because $250,000 with no matching requirement is the lowest-barrier entry into Canadian defence procurement — and the DND relationship outlasts the grant.

Not for you if: no posted challenge fits. IDEaS funds answers to its questions, not yours; there is no open call to pitch into.

Source: Department of National Defence — IDEaS Program
Closes Aug 31, 2026 Difficulty 3 / 5 $300K+ revenue required

CanExport reimburses half the cost of entering a new international market — market research, trade shows, buyer meetings, international legal and IP costs, certification — up to $50,000 per project and $99,999 per company per fiscal year. The median actual award in our catalog is about $25,000. In 2024-25, 1,575 companies were funded out of 4,406 applications.

This is the one program on this list with a revenue floor. You need $300,000 to $100 million in annual revenue, 3 to 500 full-time equivalents, incorporation, and a target market where your prior sales are under $100,000 or under 10% of total sales. Two dates bind: the 2026-27 window closes August 31, 2026 at 12:00 PM ET, and you must allow 60 business days between submitting and your first funded activity.

What to checkCanExport SMEs
AmountUp to $50K/project; $99,999/company/year; ~$25K median
Eligibility floor$300K–$100M revenue; 3–500 FTEs; incorporated
WindowFeb 4 – Aug 31, 2026, first-come competitive review
Lead time60 business days before your first activity

Verdict: The best program for a revenue-stage Canadian tech company opening a non-U.S. market is CanExport SMEs, because only $3.1 million of the $31 million 2026-27 budget is allocated to U.S.-targeted projects — meaning a CPTPP or CETA market faces far less competition per dollar. Apply before August 31 or wait until roughly February 2027.

Not for you if: you're pre-revenue or under $300,000 in annual revenue, you're a sole proprietorship or limited partnership, or your first planned activity is sooner than 60 business days out.

Source: Global Affairs Canada — CanExport SMEs
Open · runs to 2033 Difficulty 3 / 5 AI collaboration

NRC's AI for Productivity Challenge program runs from 2026 to 2033 and funds companies to build AI solutions with NRC scientists, in clean technology, agriculture and manufacturing. Contributions to industry collaborators typically fall in the $250,000 to $2 million range, negotiated per collaboration rather than posted as a fixed award. More than 250 projects were defined and initiated in the program's first year across all delivery streams, including IRAP's AI Assist channel.

Understand what you're applying for: this is a collaborative research agreement, not a cheque. The value includes access to NRC compute infrastructure, digital sandboxes and scientific staff — which for an AI startup is often worth more than the contribution. There is no posted deadline; you contact NRC to discuss current collaboration openings.

What to checkNRC AI for Productivity
Amount~$250K–$2M per collaboration (negotiated)
Application difficultyModerate (3/5)
Cost-shareNone required
WindowContinuous, program runs 2026–2033

Verdict: The best fit for an AI company already deploying models in cleantech, agriculture or manufacturing is the NRC AI for Productivity Challenge, because it bundles funding with compute and NRC scientific staff on an eight-year horizon — the only program on this list where the non-cash half is the bigger prize.

Not for you if: you want money with no strings and no partner. NRC expects to work alongside you, and the amount is negotiated rather than published.

Source: National Research Council Canada — challenge programs
Open · challenge-based Difficulty 4 / 5 Not first-timer friendly

ISC pays Canadian SMEs to solve challenges posted by federal departments: Phase 1 up to $150,000 to validate a concept, Phase 2 up to $1,000,000 to build a prototype, with the government as your first customer. Awards have averaged in the $500,000 to $600,000 range across all streams, and most Phase 1 awards land at or near the $150,000 cap.

It ranks tenth for a reason. Our catalog rates application difficulty 4 out of 5, flags the program as not first-time-applicant friendly, and marks its trend as declining: the 2025 budget cut $70 million a year from ISC, reducing the stated budget from roughly $147.6 million to about $77.6 million and, with it, the volume of new challenges. It remains a strong program — for a company that can absorb a demanding proposal and a wait.

What to checkInnovative Solutions Canada
AmountPhase 1 up to $150K; Phase 2 up to $1M (DND challenges higher)
Application difficultyHigh (4/5); not first-time-applicant friendly
Cost-shareNone — 100% funded
TrendDeclining — budget cut $70M/year from 2025-26

Verdict: The best option for a startup with a prior federal grant on its record and a technology that maps cleanly onto a posted challenge is ISC, because 100% funding with the government as pilot customer beats every other effort-to-dollar ratio here at Phase 2. Address every Essential Outcome in the challenge notice literally — evaluators cannot infer anything you did not write.

Not for you if: this would be your first government application, or you need a decision quickly — evaluation alone runs two to four months after the challenge closes.

Source: ISED — Innovative Solutions Canada

Which one to apply for first

Two branching checks that resolve most founders to a single starting program. Follow the first line that describes you.

Start here: what stage is your company at?

  • Not incorporated yet → your only routes are the ones where an institution applies. Ask a professor about SSHRC Partnership Engage (#5), and incorporate as a CCPC before doing anything else — IRAP, Mitacs and ISC all gate on it.
  • Incorporated, pre-revenue, with a research questionMitacs Accelerate (#2). Lowest difficulty, highest approval rate, ~$7,500 of your cash.
  • Incorporated, pre-revenue, doing applied AIScale AI Acceleration (#3) through a partner accelerator, then NRC AI for Productivity (#9) once you're deployed in cleantech, agriculture or manufacturing.
  • Incorporated, with a defined R&D project and technical uncertainty → open the IRAP (#1) ITA conversation now, and claim SR&ED on everything IRAP doesn't cover.
  • Above $300,000 in revenue and expanding abroadCanExport SMEs (#8), before August 31, 2026.

If more than one line fits, take the one furthest down — the later branches have higher ceilings and the earlier ones stay available year-round.

If you want an academic partner, which institution?

  • You already know a university professor in your fieldNSERC Alliance Advantage (#4). Best multiplier available: about $2 of NSERC per $1 of your cash.
  • You don't, and your problem is applied or hands-on → the applied-research office at your nearest college, via NSERC ARD (#6). Keep the first request at or under $40,000/year for the 5-to-9-week review lane.
  • Your question is about people, adoption or ethics rather than engineeringSSHRC Partnership Engage (#5), next deadline September 15, 2026.
  • You're doing quantum work and can wait for a competition → the Alliance Quantum SME Partner Stream; the current round's letter of intent closed July 27, 2026, with full applications due October 5, 2026.

In all four cases the institution is the applicant of record. Budget one to three months for finding and briefing the right researcher before any deadline you're aiming at.

AI and deep-tech programs beyond the top ten

Canada funds AI through a separate stack from general startup R&D, and most of it is sized for consortia rather than seed-stage companies. Here is the honest map — including which doors are shut today. Our wider list is at AI grants in Canada.

The four federal AI instruments an SME can actually reach

Swipe the table sideways to see every column.

ProgramSizeStatus today
Scale AI AccelerationUp to $50K in servicesOpen, rolling
NRC AI for Productivity~$250K–$2MOpen to 2033
Scale AI Supercluster projects$1M–$5M co-investmentOpen, consortium required
AI Compute Access Fund$100K–$5MBetween intakes

The take: the ladder is services → collaboration → consortium. A seed-stage AI company starts at Scale AI Acceleration, uses that cohort to find the partners a Supercluster project requires, and treats the compute fund as something to have a proposal ready for rather than something to plan around.

Scale AI Supercluster projects fund consortia, not single companies

Scale AI's industry-led project stream co-invests $1 million to $5 million per project, and the December 2025 round put $128.5 million into 44 projects — an average around $2.9 million. Our catalog puts the approval rate at 15 to 25% and difficulty at 4 out of 5, and the binding requirement is structural rather than technical: you need a consortium of two or more companies including an SME. Scale AI has broadened well past its original supply-chain remit, so applied AI in healthcare, energy, agriculture, manufacturing and public services now qualifies. The realistic path in is to attend the weekly information sessions and build the consortium first; the application is the easy part once the partners exist.

The AI Compute Access Fund is between intakes, and that matters more than its size

ACAF covers up to two-thirds of eligible Canadian cloud compute costs and half of non-Canadian compute, at $100,000 to $5 million per company — which addresses the single largest line item on most AI startups' budgets. It is not open. The first call for proposals closed July 31, 2025 and was heavily oversubscribed; our catalog records the program as between intakes with future rounds expected and no announced date. Our catalog also rates its approval rate as low, under 20%, and notes that revenue-generating companies or those with Series A or later financing were prioritised. Prepare the proposal now if compute is your bottleneck, but do not build a runway plan on a program with no posted call.

The Regional AI Initiative is a repayable contribution, not a grant

The Regional Artificial Intelligence Initiative offers $250,000 to $5 million for AI commercialization and adoption in the Prairie provinces, on continuous intake until December 31, 2028. For businesses it is an interest-free repayable contribution — you pay it back — which is why it sits outside the grant list on this page even though the amounts are large. Seven projects were approved in the program's first year, averaging about $1.07 million against a $33.8 million envelope, and the program can close early if that envelope commits. Our catalog rates it difficulty 4 and not first-time-applicant friendly. It is a genuine option for a Prairie AI company with commercialization traction; it is not free money.

How to use this list

Here is what the ranking actually encodes. Tier 1 (IRAP, Mitacs, Scale AI Acceleration) is open to any incorporated tech company with qualifying R&D and needs no third party before you can apply. Tier 2 (Alliance, Partnership Engage, ARD) requires an academic institution that will act as applicant of record — which takes one to three months to arrange but raises your ceiling permanently. Tier 3 (IDEaS, CanExport, NRC AI for Productivity, ISC) requires a specific strategic angle: a dual-use application, revenue and an export plan, deployed AI, or a posted federal challenge you happen to fit.

Here's what you need to know about stacking, which is the most important concept on this page. Canadian federal funding is deliberately layerable, and the overlap between SR&ED, IRAP, Mitacs and NSERC Alliance is the best-defined case. Each covers a different cost line on the same project: IRAP toward employee salaries, Alliance toward university-based research, Mitacs toward the graduate researcher, and SR&ED against whatever R&D spending none of them paid for — you cannot claim SR&ED on the exact dollars a grant already covered. Combined, that stack routinely offsets the majority of a project's R&D cost. Budget 2025 raised the SR&ED enhanced expenditure limit directly from $3 million to $6 million, so the maximum enhanced credit is now $2.1 million a year for a qualifying CCPC. Our IRAP vs SR&ED guide works through which to pursue first.

Treat every approval rate on this page as a base rate, not a forecast. The figures come from each program's own disclosures or our catalog data — IRAP's, for instance, is derived from 3,136 firms funded out of 9,187 reached in FY 2024-25 — and the variance around them is enormous. A well-scoped IRAP project shaped in advance with a supportive ITA has a completely different outcome distribution from the same project submitted cold. Where our catalog carries a band ("Moderate, 20-40%") rather than a number, that band is what the official documents support, and we don't sharpen it.

Two structural gates decide more outcomes than proposal quality. Incorporation: IRAP, Mitacs and ISC all require it, and CanExport additionally excludes sole proprietorships and limited partnerships outright. Who applies: on Alliance, ARD and Partnership Engage, the institution is the applicant of record — your company is the partner, so your work is sourcing the right researcher and committing cash, not writing the submission. Founders lose months to the assumption that they will be the ones filling in the form.

Tech startups · Canada · 2026

See every program you qualify for, ranked for your startup

This top ten is the shortlist. Answer a few quick questions and watch the full map narrow to the federal and provincial programs your tech company can actually get, pre-revenue included. Free, no account needed.

Questions founders ask before applying

Pre-revenue tech startups can win nine of these ten programs

Pre-revenue startups qualify for most federal funding, because Canadian programs judge R&D merit and technical capacity rather than sales. Nine of the ten programs on this page set no minimum-revenue bar. NRC IRAP funds incorporated pre-revenue firms with a defined R&D project, Mitacs Accelerate subsidizes a graduate researcher regardless of your revenue, and SR&ED refunds eligible R&D spending as a cash credit to a Canadian-controlled private corporation before you earn a dollar. The single exception here is CanExport SMEs, which requires $300,000 to $100 million in annual revenue and 3 to 500 employees. Across our whole catalog, only 15 of 452 active programs set an explicit minimum-revenue requirement — the real gates are usually incorporation, matching funds, or a program that expects a working prototype. If you're pre-revenue: incorporate as a CCPC, define one genuine R&D project, and IRAP, SR&ED and Mitacs all open at once on the same work.

IRAP, SR&ED and Mitacs stack on the same project

Stacking is the core mechanic of Canadian R&D funding, and these three are designed to layer. Each covers a different cost category on the same project: IRAP contributes toward employee salaries and contractor costs at up to 80% of eligible amounts, Mitacs Accelerate covers a graduate or postdoctoral researcher at $15,000 to $20,000 per unit, and SR&ED refunds a percentage of the remaining eligible R&D spending. A startup with a university partner can run IRAP plus NSERC Alliance plus Mitacs at once and claim SR&ED on top. The rule to remember is that you cannot claim SR&ED on the exact dollars a grant already paid — the credit applies to your unfunded R&D spending, which is why the practical strategy is to use IRAP for wages and SR&ED for the remainder. Budget 2025 raised the SR&ED enhanced expenditure limit directly from $3 million to $6 million, so the maximum enhanced credit is $2.1 million per year.

The fastest federal grants for a first-time applicant

First-time applicants win fastest with the low-barrier programs, not the flagship ones. Mitacs Accelerate and SSHRC Partnership Engage carry the lowest application difficulty on this list (2 out of 5) and both are rated "high, above 40%" for approval in our catalog; Partnership Engage runs four competitions a year, so the wait for an intake is never long. Scale AI Acceleration is the next step for AI companies at the same difficulty. Winning a smaller program first builds the organizational track record and financial-management evidence that reviewers of IRAP and NSERC Alliance look for. Two to avoid as a first application: Innovative Solutions Canada, which our catalog explicitly flags as not first-time-applicant friendly at difficulty 4, and Scale AI's Supercluster project stream, which requires a consortium. In our catalog, 26 of the 60 active federal tech-eligible programs are flagged first-time-applicant friendly, so the on-ramp is wider than most founders assume.

What "rolling intake" does and does not mean

Six of the ten programs here take applications year-round, and 36 of the 60 active federal tech-eligible programs in our catalog (60%) carry no fixed deadline at all. That is genuinely different from most provincial funding, and it is why a "grant deadline calendar" is close to useless for federal tech money. But rolling intake is not unlimited intake. IRAP's regional budgets are fullest at the start of the federal fiscal year in April and May and can be exhausted by late fall in high-demand regions, which makes February and March the right months to start an advisor relationship for the following year rather than the right months to ask for money. NSERC Alliance is continuously open but paused intake of extension requests with additional funds for existing grants in February 2025. Read "rolling" as "no deadline pressure, but real budget pressure."

What changed in 2026

Everything below moved since this list was first published in the spring. Where a change removed a program from the ranking or added one, we've said so.

Not open right now

These are good programs with no live intake as of July 31, 2026. They are worth preparing for and worth checking before you plan around them — but they are not on the ranked list, because a list of "grants you can get" should only contain grants you can currently get.

NGen SME Feasibility Studies — between intakes

Up to $100,000, covering 50% of eligible costs on a $50,000 to $200,000 feasibility study in advanced or digital manufacturing, with an estimated 40 to 60% approval rate — one of the more accessible supercluster on-ramps when it's open. It is currently listed as closed on ngen.ca. Watch ngen.ca/funding/pilot-feasibility for the next intake announcement, and note that NGen membership is free, so you can join now and be ready.

AI Compute Access Fund — between intakes

$100,000 to $5 million per company toward cloud compute, covering up to two-thirds of Canadian compute costs. The first call closed July 31, 2025 and was heavily oversubscribed; further rounds are expected but none is posted. Approval rate is recorded in our catalog as low, under 20%.

Where we track reopenings

Program status in our catalog is re-derived, not assumed — a recurring program whose dated deadline has passed flips to "between intakes" automatically rather than sitting there looking open. You can see the current state of all 452 active programs, filtered to your company, on the full catalog or through the technology grants hub.