LUMI AI Fac o y Se ice Cen e
Empowe ing Eu ope’s AI Ecosys em
D2.3 HPC and AI Ma u i y Su ey
2
D2.3
HPC and AI Ma u i y Su ey
D2.3 HPC and AI Ma u i y Su ey
3
P ojec Ti le
LUMI AI Fac o y Se ice Cen e
P ojec Ac onym
LUMI-AIF
P ojec Numbe
101234208
Type o Ac ion
HORIZON-JU-RIA
Topic
HORIZON-JU-EUROHPC-2025-AI-01-IBA-01
S a ing Da e o P ojec
01.03.2025
Ending Da e o P ojec
29.02.2028
Du a ion o he P ojec
36 mon hs
Websi e
lumi-ai- ac o y.eu
Wo k Package
WP2 Cus ome Engagemen
Task
Task 2.3. Cus ome needs analysis and acili a ing access o AI
supe compu e s
Lead Au ho s
Ou i Tasala (CSC)
Con ibu o s
Susanne Sjöblom (CSC), Ma kus Koskela (CSC), Heidi Laine
(CSC), Aleksi Kallio (CSC)
Pee Re iewe s
Ma a Maj (Cy one )
Mo hen Ma hisen (CSC)
Ve sion
1.0
Due Da e
31.7.2025
Submission Da e
28.7.2025
Dissemina ion le el
x
PU: Public
SEN: Sensi i e – limi ed unde he condi ions o he G an Ag eemen
EU-RES. Classi ied In o ma ion: RESTREINT UE (Commission Decision 2005/444/EC)
EU-CON. Classi ied In o ma ion: CONFIDENTIEL UE (Commission Decision 2005/444/EC)
EU-SEC. Classi ied In o ma ion: SECRET UE (Commission Decision 2005/444/EC)
D2.3 HPC and AI Ma u i y Su ey
4
Ve sion His o y
Re ision
Da e
Edi o s
Commen s
0.1
26.6.2025
Ou i Tasala,
Susanne Sjöblom,
Ma kus Koskela
Submi ed o in e nal e iew
0.2
1.7.2025
Ou i Tasala
Submi ed o SMB o commen s.
0.3
2.7.2025
Ou i Tasala,
Susanne Sjöblom
Sumi ed o PMO o inal quali y check.
1.0
28.7.2025
Anna Luoma
Final quali y check pe o med by he PMO,
e sion submi ed o o icial e iew.
Glossa y o Te ms
I em
Desc ip ion
Access mode
One o he a ailable ways o access he LUMI-AI supe compu e : ba ch
enan access, in e ac i e access, o exclusi e access.
Access model
One o he a ailable me hods o ge ing access o he LUMI-AI
supe compu e o di e en cus ome g oups such as SMEs, s a -ups,
and la ge indus ies: e.g. na ional and Eu opean calls, g an s,
challenges, and pay-pe -use.
Cus ome
An o ganiza ion ha uses LUMI AI Fac o y se ices.
Digi al Suppo
Sys ems
Collec ion o digi al ools used in he Cus ome P ocess, such as CRM
sys em and icke ing ool o cus ome inqui ies, des ibed in de ail in
deli e able 2.1 Cus ome P ocess Desc ip ion.
Exclusi e Access
Mode
Exclusi e access o a numbe o nodes o a gi en pe iod o ime
g an ed o a p ojec . In his way, a p ojec may ha e a p i a e clus e
wi hin LUMI-AI, wi h a ailo ed compu ing en i onmen i needed.
HPC and AI
Ma u i y
e alua ion
Ac ion ha is ei he au oma ed o ca ied ou by he AI Fac o y s a o
iden i y he en y poin s and access models o he cus ome ’s ma u i y
le el wi h HPC and AI. E alua ion is based on he su ey and he
ollowing discussion.
HPC and AI
Ma u i y su ey
A su ey ca ied ou ei he independen ly by he cus ome (online o m)
o wi h he assis ance o AI Fac o y s a (ea ly discussion) o assess he
HPC and AI eadiness as indica ion and a base o ma u i y e alua ion.
Ma u i y
Ma u i y e e s o he dep h, sophis ica ion, and in eg a ion o HPC and
AI capabili ies wi hin an o ganiza ion o sys em o e ime. Ma u i y
D2.3 HPC and AI Ma u i y Su ey
5
indica es how a he o ganiza ion has come and how well hey’ e
le e aging HPC and AI in p ac ice.
Readiness
The cu en capabili y and p epa edness o an o ganiza ion o adop
and deploy HPC and AI echnologies. Readiness indica es he s a ing
poin – how equipped he o ganiza ion is o begin o use AI o scale up
he use – and e lec s he o ganiza ions ma u i y.
Use
Pe son wi h a use accoun in LUMI o LUMI-AI, ob ained ei he om
he cus ome po al o ia he Eu oHPC Fede a ion Pla o m.
D2.3 HPC and AI Ma u i y Su ey
6
Execu i e Summa y
This epo p esen s he HPC and AI Ma u i y su ey de eloped by he LUMI AI Fac o y as a s a egic
ool o suppo cus ome s in assessing hei po en ial o adop high-pe o mance compu ing (HPC) and
a i icial in elligence (AI) echnologies. The su ey plays a cen al ole in guiding cus ome s h ough he
p ocess o de ining p ojec scope, unde s anding a ailable esou ces, and p epa ing success ul
applica ions o compu e capaci y h ough LUMI AI Fac o y and Eu oHPC JU.
The p ima y objec i e o he su ey is o e alua e an o ganiza ion’s cu en capabili ies, iden i y
echnical and skills gaps, and ecommend sui able en y poin s and access models o compu e
esou ces. I is an in eg al pa o he AI Fac o y’s cus ome p ocess, ensu ing ha cus ome s a e well-
in o med and posi ioned o success in le e aging ad anced compu ing echnologies.
In addi ion o i s di ec bene i s o indi idual cus ome s, he su ey enables LUMI AI Fac o y o collec
and analyze da a ac oss o ganiza ions, unco e ing b oade ends and in o ming u u e suppo
s a egies. This dual unc ion enhances bo h indi idual p ojec ou comes and he o e all e ec i eness
o AI Fac o y’s se ices.
Looking ahead, he epo ou lines a oadmap o e iewing and de eloping he su ey. Key
ecommenda ions include inco po a ing use eedback, modula izing he su ey o domain-speci ic
ele ance, and in eg a ing i wi h AI Fac o y’s digi al suppo sys ems. These enhancemen s will ensu e
he su ey emains a dynamic, use -cen e ed ool ha e ol es alongside echnological ad ancemen s
and cus ome needs.
D2.3 HPC and AI Ma u i y Su ey
7
Table o Con en s
1. In oduc ion ............................................................................................................... 8
2. Objec i e .................................................................................................................... 8
3. Scope ....................................................................................................................... 10
4. Me hod o eaching cus ome s ................................................................................ 11
5. C i e ia o sco ing .................................................................................................... 12
6. HPC and AI Ma u i y Su ey empla e s uc u e ........................................................ 13
A) Ques ions o es ima e he eadiness le el 14
B) Ques ions o es ima e he need ca ego y 19
7. Requi emen s o he su ey ool used o HPC and AI Ma u i y e alua ion ............... 22
8. Fu u e e iew and de elopmen o he HPC and AI Ma u i y Su ey .......................... 23
9. Conclusions .............................................................................................................. 23
D2.3 HPC and AI Ma u i y Su ey
8
1. In oduc ion
This documen desc ibes he comp ehensi e suppo ac ions by he LUMI AI Fac o y o cus ome s
applying o conside ing applying o HPC and AI compu ing esou ces, ensu ing success ul p ojec
acquisi ion and cus ome sa is ac ion. I also p o ides a classi ica ion o HPC and AI eadiness and
in oduces a empla e o he HPC and AI Ma u i y Su ey, he main ool used in he e alua ion. HPC
and AI Ma u i y Su ey helps he cus ome in de ining he p ojec scope, objec i es, and imelines, as
well as helping hem unde s and and na iga e he a ailable HPC and AI esou ces, ools, and se ices.
Fo comple ing he e alua ion, he su ey is supplemen ed by an in-dep h discussion.
This deli e able in oduces he empla e used o he su ey and he me hods o assessing cus ome ’s
HPC and AI ma u i y. I also makes explici he planned ac ions o de eloping he HPC and AI Ma u i y
Su ey and he analysis based on i u he .
2. Objec i e
The objec i e o he HPC and AI Ma u i y e alua ion is o unde s and cus ome s’ cu en capabili ies
o le e aging AI and HPC echnologies o o m a cohe en pic u e o he cus ome s’ ma u i y o
hemsel es and o he LUMI AI Fac o y. The analysis does his by co e ing key dimensions ela ed o
HPC capabili ies, access and owne ship o da ase s and expe iences wi h AI applica ions. I desc ibes
he o ganiza ion’s echnical eadiness o AI and HPC de elopmen and shows possible skills gaps and
compe ence needs. Readiness is used he e o e lec he o ganiza ion's cu en capabili y o adop and
scale HPC and AI echnologies, se ing as a p ac ical indica o o he o ganiza ions o e all digi al
ma u i y in HPC. The main pu pose o he HPC and AI ma u i y e alua ion is o help he cus ome
na iga e wi hin he compu e se ices a ailable h ough LUMI AI Fac o y and Eu oHPC JU.
HPC and AI ma u i y e alua ion is an in eg al pa o he cus ome p ocess in LUMI AI Fac o y.
Cus ome jou ney is desc ibed in mo e de ail in deli e able 2.1 Cus ome P ocess Desc ip ion and only
he simpli ied g aph o he cus ome jou ney is p esen ed he e o e e ence:
D2.3 HPC and AI Ma u i y Su ey
9
Figu e 1 shows a simpli ied cus ome jou ney wi h he di e en s eps along he pa h, he LUMI AI Fac o y se ices and he
cus ome needs analysis and he HPC and AI ma u i y analysis as pa o he p ocess.
The HPC and AI ma u i y e alua ion aims o:
• Implemen he unde s anding o he cus ome ’s cu en s a e and need, gained om he
Cus ome Needs Analysis
• De e mine he cus ome ’s po en ial o apply o esou ces
I he esul s show ha he p ojec needs compu e esou ces, he aim is o:
• Ge he cus ome acquain ed wi h he equi emen s o applying o esou ces and he di e en
access models
• Aid he cus ome in w i ing a success ul p oposal in applying o HPC esou ces sui able o hei
needs
The objec i e o HPC and AI ma u i y su ey is o gi e cus ome s an in o med pic u e o hei HPC and
AI ma u i y and aid hem on hei jou ney owa ds ge ing access o compu e esou ces. The
assessmen p o ides ecommenda ions o he mos sui able en y poin and access model o ha
cus ome o apply o HPC and AI capaci y. The aspec s included in he su ey a e ele an in he
applica ion p ocess o he esou ces, and hus his phase in he cus ome p ocess p epa es he
cus ome o applying o he esou ces.
D2.3 HPC and AI Ma u i y Su ey
16
Da a collec ion
and p epa a ion
2
Model selec ion
and de elopmen
3
Deploymen
3
Ha e you p e iously p o en in some en i onmen ha he applica ion o code is scalable?
Sco e ange: 0-3. Weigh : 3.
Response
Sco e
No scalabili y es ing o alida ion has been done.
0
Limi ed scalabili y es ing; he applica ion has been un on a small clus e o wi h a
modes numbe o co es/nodes, bu no o mal scalabili y analysis.
1
Mode a e scalabili y p o en; he applica ion has been es ed on mid-sized HPC
sys ems wi h some pe o mance me ics o scaling beha io obse ed.
2
Scalabili y has been ho oughly demons a ed; he applica ion has been benchma ked
o p o iled on la ge-scale HPC sys ems, wi h documen ed s ong scaling o weak
scaling esul s.
3
Wha ypes o wo kloads ha e you un on HPC sys ems (e.g., simula ions, AI/ML aining,
da a analy ics)? Sco e ange: 0-3. Weigh : 2.
Response
Sco e
No HPC wo kloads un ye .
0
Basic o single-domain wo kloads (e.g., simple simula ions, ba ch p ocessing, o basic
da a analysis).
1
Mode a e di e si y o complexi y (e.g., mul i-domain simula ions, AI/ML aining, o
la ge-scale da a analy ics).
2
Ad anced and a ied wo kloads (e.g., igh ly coupled simula ions, hyb id AI/HPC
wo k lows, eal- ime da a p ocessing, o wo k lows equi ing GPUs, high memo y, o
I/O op imiza ion).
3
Wha job schedule s o esou ce manage s a e you amilia wi h (e.g., Slu m, PBS, LSF)?
Sco e ange: 0-3. Weigh : 3.
D2.3 HPC and AI Ma u i y Su ey
17
Response
Sco e
No amilia i y wi h job schedule s o esou ce manage s.
0
Basic awa eness o limi ed expe ience (e.g., submi ed jobs using a empla e bu no
amilia wi h schedule op ions o commands).
1
Com o able using one o mo e schedule s (e.g., Slu m, PBS, LSF); can w i e and
modi y job sc ip s, use job a ays, and moni o jobs.
2
Ad anced p o iciency; expe ienced wi h mul iple schedule s, unde s ands esou ce
alloca ion, job dependencies, and can op imize job scheduling o manage queues.
3
A e you applica ions op imized o pa allel o dis ibu ed compu ing? Sco e ange: 0-3.
Weigh : 3.
Response
Sco e
No op imiza ion o pa allel o dis ibu ed compu ing; applica ions un se ially o on a
single co e.
0
Basic pa alleliza ion implemen ed (e.g., mul i h eading), bu limi ed scalabili y o
e iciency.
1
Applica ions a e easonably op imized o pa allel o dis ibu ed en i onmen s;
demons a e good pe o mance on mode a e co e coun s o nodes.
2
Applica ions a e highly op imized o HPC; include ad anced pa alleliza ion s a egies
(e.g., 3D pa allelism) e icien memo y and I/O usage, and p o en scalabili y on la ge
sys ems.
3
Do you use con aine s o wo k low manage s (e.g., Singula i y, Nex low, Snakemake)?
Sco e ange: 0-3. Weigh : 3.
Response
Sco e
No use o con aine s o wo k low manage s; all wo k is done manually o in ad hoc
sc ip s.
0
Basic awa eness o occasional use o con aine s o wo k low ools, bu no in eg a ed
in o egula wo k lows.
1
D2.3 HPC and AI Ma u i y Su ey
18
Regula use o con aine s (e.g., Singula i y, Docke ) o wo k low manage s (e.g.,
Snakemake, Nex low) o imp o e ep oducibili y and au oma ion.
2
Ad anced use o con aine s and wo k low manage s; wo k lows a e ully con aine ized,
po able, and au oma ed, wi h suppo o scalabili y, ep oducibili y, and e sion
con ol.
3
A e you using any e sion con ol o CI/CD ools in you wo k lows? Sco e ange: 0-3.
Weigh : 2.
Response
Sco e
No e sion con ol o CI/CD ools a e used; code and wo k lows a e managed manually.
0
Basic use o e sion con ol (e.g., Gi ) o code acking, bu no s uc u ed wo k low o
au oma ion.
1
Regula use o e sion con ol and some CI/CD p ac ices (e.g., au oma ed es ing,
deploymen sc ip s, Gi Hub Ac ions).
2
Fully in eg a ed e sion con ol and CI/CD pipelines; includes au oma ed es ing,
deploymen , documen a ion, and ep oducibili y ac oss en i onmen s.
3
Do you ha e a dedica ed eam o pe son managing HPC esou ces? Sco e ange: 0-3.
Weigh : 3.
Response
Sco e
No dedica ed pe son o eam
0
One pe son has pa ial esponsibili y o HPC asks, bu i 's no hei p ima y ole.
1
A dedica ed pe son manages HPC esou ces, suppo s use s, and handles basic
main enance o job oubleshoo ing.
2
A dedica ed HPC eam is in place, wi h clea oles o sys em adminis a ion, use
suppo , pe o mance uning, and wo k low op imiza ion.
3
Do you hold he necessa y da a you need o you analysis? Sco e ange: 0-3. Weigh : 3.
Response
Sco e
D2.3 HPC and AI Ma u i y Su ey
19
No ele an da a is cu en ly a ailable; da a collec ion has no s a ed o is a majo
blocke .
0
Some da a is a ailable, bu i is incomple e, ou da ed, o no in a usable o ma o he
quali y o he da a is unknown.
1
Mos o he equi ed da a is a ailable and usable, hough some gaps o quali y issues
may emain.
2
All necessa y da a is a ailable, well-o ganized, and eady o analysis; includes p ope
documen a ion, o ma s, and access con ols.
3
Do you ha e a da a managemen o backup s a egy in place? Sco e ange: 0-3. Weigh : 2.
Response
Sco e
No da a managemen o backup s a egy.
0
Basic s a egy in place.
1
A consis en da a managemen and backup p ocess exis s; includes e sioning, egula
backups, and some documen a ion.
2
A comp ehensi e and au oma ed da a managemen s a egy is in place; includes
egula backups, me ada a, access con ol, da a li ecycle policies, and compliance wi h
bes p ac ices o s anda ds.
3
B) Ques ions o es ima e he need ca ego y
How much compu ing capaci y does you wo k equi e? Sco e ange: 0-3. Weigh : 1.
Response
Sco e
None
0
Small scale <5000 GPU hou s
1
Medium scale, 10 000 - 50 000 GPU hou s
2
La ge scale, > 50 000 GPU hou s
3
How long do you jobs usually un? Sco e ange: 0-3. Weigh : 1.
D2.3 HPC and AI Ma u i y Su ey
20
Response
Sco e
Jobs a e e y sho (e.g., a ew minu es); ypically es uns o ligh weigh asks.
0
Jobs un o up o a ew hou s.
1
Jobs ypically un o se e al hou s o a day.
2
Jobs egula ly un o mul iple days o equi e checkpoin ing.
3
Do you an icipa e scaling you wo kloads signi ican ly in he nea u u e? Sco e ange: 0-3.
Weigh : 1.
Response
Sco e
No plans o scale wo kloads; cu en usage is expec ed o emain he same.
0
Mino scaling an icipa ed (e.g., sligh ly la ge da ase s o mo e equen uns).
1
Mode a e scaling expec ed (e.g., new p ojec s, inc eased da a olume, o mo e use s).
2
Signi ican scaling planned (e.g., majo inc ease in compu e demand, ansi ion o
la ge-scale simula ions, AI/ML model aining, o mul i-node wo k lows).
3
Do you ha e you own applica ion o sel -de eloped codebase? Sco e ange: 0-3. Weigh : 1.
• Own applica ion e e s o a comple e so wa e p oduc ha you o you o ganiza ion
has c ea ed, a inished p oduc .
• Sel -de eloped codebase e e s o code w i en by you o you eam ha may be pa
o an applica ion, lib a y, o ool.
Response
Sco e
No sel -de eloped code; only uses hi d-pa y applica ion o ools.
0
Mino modi ica ions o exis ing codebases o sc ip s; limi ed o iginal de elopmen .
1
Cus om sc ip s o applica ions ailo ed o speci ic asks o wo k lows.
2
Fully sel -de eloped, complex codebase o applica ion; ac i ely de eloped, e sion-
con olled, and op imized o HPC use.
3
Does you company use open-sou ce o p op ie a y so wa e o AI and HPC? Sco e ange:
0-3. Weigh : 1.
D2.3 HPC and AI Ma u i y Su ey
21
• P op ie a y so wa e is owned by an indi idual o company. Sou ce code is closed o he
public.
• Open-sou ce so wa e is eely a ailable o anyone o use, modi y, and dis ibu e.
Sou ce code is open and accessible.
Response
Sco e
No clea s a egy; limi ed o no use o specialized so wa e o AI o HPC.
0
Uses only p op ie a y o only open-sou ce so wa e, wi h limi ed lexibili y o
in eg a ion.
1
Uses a mix o open-sou ce and p op ie a y ools, bu wi h limi ed cus omiza ion o
op imiza ion.
2
Ac i ely uses and in eg a es bo h open-sou ce and p op ie a y so wa e s a egically;
con ibu es o open-sou ce p ojec s o cus omizes ools o pe o mance and
scalabili y.
3
Does you wo k possess/gene a e/use la ge da ase s? Sco e ange: 0-3. Weigh : 1.
Response
Sco e
No o < 10 GB
0
Small-scale, 10 GBs - 100 GB
1
Medium-scale, 100 GBs - 1 TB
2
La ge-scale, > 1 TBs
3
Is he da a in he p ojec sensi i e o con iden ial? Sco e ange: 0-3. Weigh : 1.
• Sensi i e da a is such ha , i exposed, could cause ha m o indi iduals, o ganiza ions,
o sys ems. This includes pe sonal da a, heal h eco ds, inancial de ails, o p op ie a y
business in o ma ion.
• Con iden ial da a is es ic ed o au ho ized use s only, o en p o ec ed by legal,
con ac ual, o o ganiza ional policies. Disclosu e could iola e ag eemen s o
egula ions.
Response
Sco e
No
0
D2.3 HPC and AI Ma u i y Su ey
22
Yes, con iden ial
2
Yes, sensi i e
3
Is he da a in he p ojec dynamic o s eaming? Sco e ange: 0-3. Weigh : 1.
• Dynamic da a changes o e ime, i can be upda ed, modi ied, o eplaced pe iodically.
• S eaming da a is con inuously gene a ed and deli e ed in eal ime o nea - eal ime.
Da a is s eamed in o he en i onmen om he ou side.
Response
Sco e
No
0
Yes, dynamic
2
Yes, s eaming
3
7. Requi emen s o he su ey ool used o HPC and
AI Ma u i y e alua ion
A he beginning, a simple su ey ool, such as Type o m o simila , is implemen ed o ge he ma u i y
su eys unning. Howe e , some needs ha e been iden i ied ha demand u he in es iga ion in o he
selec ion o he ool.
To enhance he unc ionali y and use expe ience o he su ey ool, i should include a mechanism o
sco ing use esponses in he backg ound. This sco ing p ocess mus be au oma ed and in isible o he
use , allowing o eal- ime analysis o inpu da a. Based on he calcula ed sco es, he sys em should be
capable o gene a ing pe sonalized sugges ions o ecommenda ions ha a e ele an o he use 's
esponses. This ea u e aims o p o ide immedia e, con ex -awa e eedback, he eby inc easing he
ool’s in e ac i i y and alue.
Requi emen s:
• The sys em shall assign sco es o use esponses using p ede ined logic o algo i hms.
• The sco ing p ocess shall be execu ed in he backg ound wi hou equi ing use in e en ion.
• The sys em shall use he sco es o gene a e con ex - ele an sugges ions o ecommenda ions.
• The sugges ions shall be p esen ed o he use in a clea and ac ionable o ma .
• The sco ing logic shall be con igu able o suppo di e en su ey ypes o use cases.
D2.3 HPC and AI Ma u i y Su ey
23
8. Fu u e e iew and de elopmen o he HPC and AI
Ma u i y Su ey
To ensu e he con inued ele ance and e ec i eness o he HPC and AI Ma u i y Su ey, a egula
e iew cycle o he su ey is es ablished. This includes eedback collec ion om bo h in e nal
s akeholde s and cus ome s who ha e comple ed he su ey. Thei insigh s can highligh a eas whe e
he su ey may be oo complex, oo simplis ic, o missing c i ical dimensions o HPC and AI eadiness.
Inco po a ing his eedback will help e ine he ques ions, imp o e cla i y, and ensu e he su ey
emains aligned wi h e ol ing echnologies and cus ome needs.
Ano he a ea o de elopmen is modula iza ion and cus omiza ion o he su ey. As AI and HPC use
cases di e si y, a one-size- i s-all app oach is no desi ed. De eloping modula sec ions ailo ed o
speci ic domains (e.g., heal hca e, manu ac u ing, academia) o o ganiza ion ypes could make he
su ey mo e ele an and ac ionable. This cus omiza ion could be suppo ed by an adap i e digi al
in e ace ha adjus s he su ey low based on ini ial esponses.
Ano he a enue o u u e de elopmen is he in eg a ion wi h AI Fac o y’s b oade cus ome suppo
sys ems. In eg a ion wi h CRM sys em o p ojec acking ools would enable a seamless expe ience
om ini ial assessmen o esou ce applica ion and p ojec execu ion.
9. Conclusions
The HPC and AI ma u i y su ey is designed o be a aluable ool in suppo ing cus ome s on hei
jou ney owa d e ec i ely u ilizing high-pe o mance compu ing and AI esou ces. By assessing an
o ganiza ion’s cu en capabili ies, echnical eadiness, and po en ial skills gaps, he su ey enables
ailo ed guidance and s a egic alignmen wi h a ailable compu e se ices. I no only acili a es
success ul p ojec planning and p oposal de elopmen bu also enhances he o e all cus ome
expe ience wi hin he AI Fac o y amewo k.
Fu he mo e, he agg ega ed insigh s om he su ey con ibu e o a b oade unde s anding o ends
ac oss indus ies and domains, allowing AI Fac o y o con inuously e ine i s suppo mechanisms. As
he su ey e ol es, i will play an inc easingly c i ical ole in empowe ing cus ome s o make in o med
decisions and maximize he impac o HPC and AI echnologies in hei ope a ions.
