New be efit builds o DeepI f a’s expa ded DeepSta t p og am fo ea ly-stage AI compa ies
PALO ALTO, Calif., Nov. 03, 2026 (GLOBE NEWSWIRE) — DeepI f a, a pu pose-built cloud platfo m fo high-th oughput AI i fe e ce, today a ou ced a ew be efit fo membe s of the NVIDIA I ceptio p og am, p ovidi g eligible sta tups with 10 billio cached-i put toke s fo the NVIDIA Nemot o family models o DeepI f a’s i fe e ce platfo m. At the same time, the compa y is also expa di g DeepSta t, its p og am fo ea ly-stage AI compa ies, to p ovide 50 billio f ee toke s.
“AI sta tups a e movi g f om p ototype to p oductio faste tha eve , a d i f ast uctu e ca quickly become the co st ai t,” said Nikola Bo isov, co-fou de a d CEO of DeepI f a. “We wa t teams to be able to focus o buildi g p oducts i stead of ma agi g i f ast uctu e. By expa di g DeepSta t a d wo ki g with NVIDIA I ceptio , we’ e givi g mo e compa ies the capacity to test, build a d scale o the same i f ast uctu e they ca use i p oductio .”
The ew NVIDIA I ceptio be efit gives sta tups access to NVIDIA Nemot o family models th ough DeepI f a’s se ve less i fe e ce platfo m, with o i f ast uctu e to p ovisio o ma age a d p oductio -g ade th oughput o NVIDIA’s latest-ge e atio accele ated computi g i f ast uctu e. NVIDIA I ceptio membe s accepted i to the p og am ca also take adva tage of DeepSta t, givi g them additio al esou ces to use ac oss models available o DeepI f a.
DeepI f a c eated DeepSta t to give ea ly-stage AI compa ies access to p oductio -g ade i fe e ce i f ast uctu e as they develop a d scale ew applicatio s. The expa ded p og am ow p ovides qualifyi g sta tups with app oximately 50 billio f ee toke s, up f om the p og am’s p evious offe i g of 1 billio f ee toke s. C edits a e available fo o e yea followi g app oval.
The p og ams come as AI applicatio s i c easi gly move i to p oductio , whe e i fe e ce cost, late cy, a d eliability become c itical co st ai ts. DeepI f a’s ve tically i teg ated platfo m combi es dedicated AI compute, optimized softwa e, a d a fully ma aged develope expe ie ce, allowi g compa ies to u mo e tha 200 models th ough Ope AI-compatible APIs without havi g to build o ma age thei ow GPU i f ast uctu e.
DeepI f a wo ks closely with NVIDIA ac oss its i f ast uctu e a d model ecosystem: it has fi st-pa ty i teg atio with NVIDIA Ope Shell, a d DeepI f a was pa t of a select g oup of collabo ato s to u a i depe de t be chma k o NVIDIA Ve a CPU.
Compa ies i cludi g Ope Code, LiveKit a d huma s& have used DeepSta t to build a d scale AI p oducts o DeepI f a.
NVIDIA I ceptio membe s ca lea mo e about the be efit at deepi f a.com/ vidia-i ceptio .
Sta tups ca lea mo e about DeepSta t at deepi f a.com/deepsta t.
Fou ded i 2022, DeepI f a is a pu pose-built cloud i fe e ce platfo m fo high-th oughput AI, e abli g compa ies to u ope -sou ce a d p op ieta y AI a d age tic models at scale. The compa y ow s a d ope ates its AI Compute i f ast uctu e, suppo ts 200+ ope -sou ce models, a d p ocesses ove 22 t illio toke s pe week — delive i g the cost efficie cy, low late cy, a d secu ity that p oductio AI dema ds. Its fully ma aged platfo m i cludes Ope AI-compatible APIs a d e te p ise-g ade secu ity complia ce, i cludi g ze o data ete tio a d SOC 2 a d ISO 27001 ce tificatio . Fo mo e i fo matio , visit deepi f a.com.
