HONG KONG, July 20, 2026 (GLOBE NEWSWIRE) — 3 E Netwo k Tech ology G oup Limited (Nasdaq: MASK) (the “Compa y” o “3 E Netwo k”), a busi ess-to-busi ess (“B2B”) i fo matio tech ology (“IT”) busi ess solutio s p ovide , committed to becomi g a ext-ge e atio a tificial i tellige ce (“AI”) i f ast uctu e solutio s p ovide , today a ou ced a ew developme t i its semico ducto busi ess segme t. Followi g the st ategic f amewo k ag eeme t e te ed i to with Califo ia-based adva ced obotics e te p ise Aladdi Ala is AI I c. (“Aladdi Ala is AI”), the Compa y a ou ced that it has completed key a chitectu al desig wo k a d edge computi g deployme t pla i g fo its custom Edge AI System-o -Chip (“SoC”), which is bei g desig ed fo ext-ge e atio sma t healthca e a d elde ca e obots.
Compa ed to t aditio al i dust ial obotic a ms physically isolated by “safety fe ces” i st uctu ed facto ies, the computatio al challe ges faci g sma t elde ca e obots ep ese t a sig ifica t i c ease i complexity. They may ope ate without the p otective ba ie s commo ly used i i dust ial setti gs a d must avigate dy amic, highly u st uctu ed eal-wo ld home e vi o me ts to e gage i f eque t, di ect physical co tact with huma s, pa ticula ly olde adults who may equi e additio al ca e. This shift f om isolated e vi o me ts to high-f eque cy physical huma – obot i te actio imposes th ee st ict tech ical equi eme ts o the u de lyi g compute powe :
High Real-Time Respo sive ess a d Physical Safety: Whe faci g sudde eme ge cies like a elde ly pe so falli g, o whe maki g physical co tact to assist with sta di g, the obot must p ocess massive amou ts of multi-modal se so data co cu e tly with ult a-low late cy. Eve hu d eds of milliseco ds of cloud etwo k late cy could esult i se ious safety haza ds.St ict Complia ce Requi eme ts fo Medical P ivacy: To add ess applicable medical p ivacy equi eme ts, i cludi g those that may a ise u de the U.S. Health I su a ce Po tability a d Accou tability Act of 1996 (“HIPAA”), high-defi itio video, audio footage, a d vital sig data captu ed co ti uously i the home may equi e app op iate safegua ds befo e bei g t a smitted to o p ocessed i the cloud. This ma dates that the obot possess exceptio ally obust “local closed-loop” computi g capabilities.Netwo k Bli d Spots a d Offli e Reliability: Netwo k bli d spots a d sig al fluctuatio s f eque tly occu i complex a d dy amic home e vi o me ts. Howeve , life-savi g p otectio a d atu al emotio al compa io ship ca ot be co ti ge t upo u i te upted i te et co ectivity. I the eve t of a etwo k outage, elde ca e obots must possess a obust local computi g fou datio capable of i depe de tly executi g adva ced cog itive a d ca egivi g st ategies.
Faced with these complex, u st uctu ed ope ati g co ditio s, t aditio al me cha t silico p io itizi g ge e alized be chma k met ics ofte falls sho t due to a chitectu al limitatio s, suffe i g f om late cy, powe , a d ba dwidth bottle ecks. This mismatch i u de lyi g compute has co t ibuted to the shift f om ge e al-pu pose silico to “sce a io-defi ed” semico ducto desig .
Co seque tly, 3 E Netwo k’s custom-developed Edge AI SoC is focused o the optimizatio of edge i fe e ce capabilities. Utilizi g a “ha dwa e-softwa e co-desig ” st ategy, the chip a chitectu e p io itizes low late cy a d e e gy efficie cy fo specific physical i te actio s, systematically add essi g the u de lyi g computi g challe ges of elde ca e sce a ios:
Milliseco d-Level Multi-Modal Hete oge eous Fusio Pe ceptio : The custom chip utilizes a highly optimized hete oge eous computi g a chitectu e. By i teg ati g ha dwa e accele atio e gi es desig ed specifically fo complex physical i te actio s, the chip e ables the pa allel scheduli g of high-f ame- ate depth visio (RGB-D), millimete -wave spatial ada , a d complia t tactile a ay data di ectly at the silico level. This empowe s the obot to complete 3D spatial modeli g a d dy amic obstacle avoida ce with milliseco d-level late cy, eve i offli e co ditio s.Ha dwa e-Level T usted Executio E vi o me t (“TEE”) a d P ivacy Isolatio : Add essi g st ict p ivacy complia ce equi eme ts, the chip establishes a secu e TEE at the ha dwa e level. The a chitectu e is desig ed to e able home visual footage, voice comma ds a d vital sig data to be de-ide tified a d p ocessed fo featu e ext actio withi this e c ypted ha dwa e sa dbox. This e su es that co e data is p ocessed st ictly o -device, helpi g educe p ivacy a d complia ce isks associated with cloud data t a smissio .Localized Qua tized La ge Model LLM/VLM I fe e ce E gi e: This desig st ategy is co siste t with the b oade i dust y t e d towa d o -device p ocessi g a d p ivacy-p ese vi g computi g. The chip’s a chitectu e is specifically optimized fo la ge models. Leve agi g exceptio al te so compute-to-powe efficie cy, obots ca smoothly execute qua tized local La ge La guage Models a d Visio -La guage Models at the edge, e abli g atu al sema tic u de sta di g, emotio ecog itio , a d daily compa io co ve satio s.Ult a-Low Late cy Tactile Feedback a d Huma -Like Complia t Co t ol: Ta geti g the co e safety challe ge of physical huma – obot co tact, the Edge AI SoC ese ves a dedicated ult a-low-late cy co t ol data path specifically fo p ocessi g fi e-g ai ed feedback f om highly se sitive complia t tactile a d to que se so s i eal time. The u de lyi g compute ca d ive obotic a ms at kilohe tz ef esh ates to achieve highly complia t, “huma -like” fo ce co t ol, sig ifica tly mi imizi g safety isks caused by igid mecha ical collisio s.Milliwatt-Level “Always-O ” Dy amic Powe A chitectu e: Elde ca e obots equi e 24/7 co ti uous mo ito i g. The chip i t oduces a adva ced dy amic powe ma ageme t mecha ism. I outi e sta dby mode, it ope ates at a powe level of o ly a few milliwatts to mai tai specific wake-wo d liste i g a d c itical eve t detectio , such as fall detectio . O ce a th eshold is t igge ed, the chip ca apidly activate the system’s full computi g capacity, p ovidi g ou d-the-clock safety edu da cy.
U like t aditio al ge e al-pu pose silico that attempts a “o e-size-fits-all” desig app oach, 3 E Netwo k’s Edge AI SoC p ofou dly p actices the comme cial philosophy of “Sce a io-Defi ed Silico .” Th ough deep co-desig with Aladdi Ala is AI du i g the ea ly a chitectu al stages, the chip’s u de lyi g ha dwa e a chitectu e a d eal-time co t ol algo ithms have bee deeply customized a d optimized fo complex physical i te actio o ms. Alig i g with the specialized computi g t e d highlighted by i dust y-leadi g platfo ms like NVIDIA’s Jetso Tho , 3 E Netwo k is actively buildi g its tech ological moat th ough this highly ve tical custom SoC, accele ati g the time-to-ma ket fo mass p oductio .
D . Ti gju Ya g, Chief Executive Office of 3 E Netwo k, stated: “I the e a of Embodied AI, the co e tech ological ba ie lies i the capability of eal-time physical i te actio at the edge. The Edge AI SoC we a e custom-buildi g fo Aladdi ’s elde ca e obots aims to p ovide obust edge computi g a d espo sive ess fo the obots. By empowe i g obots with outsta di g local multi-modal pe ceptio a d a high deg ee of data secu ity, we a e t a slati g silico -level compute powe i to the fou datio of sma t ca egivi g that ta gibly be efits the agi g populatio . This ma ks a impo ta t step i the developme t a d pote tial comme cializatio of 3 E Netwo k’s semico ducto busi ess.”
This developme t eflects the Compa y’s effo ts to t a slate its semico ducto expe tise i to custom silico solutio s fo embodied AI applicatio s. As the R&D p ocess fo this customized Edge AI SoC steadily adva ces, the Compa y is actively d ivi g towa ds tape-out a d subseque t validatio milesto es, co ti ui g to co solidate its competitive adva tage i the u de lyi g compute ma ket fo global se vice obots.
