The world is un­der­go­ing a fun­da­men­tal trans­for­ma­tion. The ques­tion is whether T&T will keep pace with this rate of change. If we do not, then we will be falling be­hind. Take the com­put­er, a de­vice that we use every­day. For 60 years, it has op­er­at­ed un­der the same ba­sic mod­el: pro­gram­mers write ex­plic­it in­struc­tions, proces­sors ex­e­cute them se­quen­tial­ly, and ap­pli­ca­tions re­trieve pre­built re­spons­es from stor­age. That mod­el is end­ing. In its place, a new par­a­digm is emerg­ing that will make the close of 2026 look re­mark­ably dif­fer­ent from where we stand to­day. For T&T, an econ­o­my al­ready grap­pling with the en­er­gy tran­si­tion and mount­ing de­mo­graph­ic pres­sures, this shift presents both sig­nif­i­cant chal­lenges and un­ex­pect­ed op­por­tu­ni­ties that de­mand our im­me­di­ate at­ten­tion.

Jensen Huang, the chief ex­ec­u­tive of Nvidia, re­cent­ly ar­tic­u­lat­ed this trans­for­ma­tion: “The en­tire five lay­er stack of the com­put­er in­dus­try is be­ing rein­vent­ed. You no longer pro­gram the soft­ware, you train the soft­ware. You don’t run it on CPUs, (cen­tral pro­cess­ing unit) you run it on GPUs (graph­ics pro­cess­ing unit). And where­as ap­pli­ca­tions were pre record­ed, pre com­piled, and run on your de­vice, now ap­pli­ca­tions un­der­stand the con­text and gen­er­ate every sin­gle pix­el, every sin­gle to­ken, com­plete­ly from scratch every sin­gle time.”

This is not hy­per­bole from a man sell­ing graph­ics proces­sors. It is a pre­cise de­scrip­tion of the in­fra­struc­tur­al rev­o­lu­tion now un­der­way. When you queried a tra­di­tion­al search en­gine, it re­trieved in­dexed web pages. When you ask ar­ti­fi­cial in­tel­li­gence the same ques­tion, it gen­er­ates a unique re­sponse tai­lored to your spe­cif­ic con­text. Every word, every out­put, cre­at­ed anew rather than pulled from stor­age. The im­pli­ca­tions for cap­i­tal al­lo­ca­tion, labour mar­kets and eco­nom­ic pro­duc­tiv­i­ty are pro­found, and small open economies like ours can­not af­ford to re­main pas­sive ob­servers while the rules of the glob­al econ­o­my are be­ing rewrit­ten.

Con­ver­gence

The year 2026 will be re­mem­bered as the mo­ment two par­al­lel tracks of ar­ti­fi­cial in­tel­li­gence de­vel­op­ment be­gan to con­verge. The first track, large lan­guage mod­el-based gen­er­a­tive AI, has dom­i­nat­ed cap­i­tal flows and head­lines since Chat­G­PT’s re­lease in No­vem­ber 2022. The sec­ond track, em­bod­ied AI or ro­bot­ics, has pro­ceed­ed more qui­et­ly. Since the re­lease of Chat­G­PT, in­vestors have large­ly fo­cused on gen­er­a­tive AI. By con­trast, the era of hu­manoid ro­bots still feels dis­tant to most ob­servers sit­ting here in T&T.

But per­cep­tions about these two tra­jec­to­ries may soon flip. Ar­ti­fi­cial gen­er­al in­tel­li­gence may be much fur­ther out than its evan­ge­lists claim, and large lan­guage mod­els might not even be the right par­a­digm to reach it. Mean­while hu­manoid and pur­pose spe­cif­ic ro­bots, from self-dri­ving sys­tems to an­thro­po­mor­phic as­sis­tants, may be­come re­al­i­ty soon­er than ex­pect­ed. The crit­i­cal in­sight is that ro­bots do not need hu­man-lev­el in­tel­li­gence to be im­mense­ly use­ful, es­pe­cial­ly in an age­ing world des­per­ate for so­lu­tions to labour short­ages. T&T, with its own de­mo­graph­ic tran­si­tion ac­cel­er­at­ing as birth rates de­cline and the pop­u­la­tion ages, should pay par­tic­u­lar at­ten­tion to these de­vel­op­ments.

The age of ro­bot­ics with mul­ti­modal large lan­guage mod­el brains is fi­nal­ly ar­riv­ing. These ma­chines will un­der­stand in­struc­tions and per­form tasks. They can use a lap­top, wash dish­es and make cof­fee. The time­line pro­jec­tions from those clos­est to the tech­nol­o­gy sug­gest five to ten times more ro­bot­ics com­pa­nies will be found­ed in the next 12 months. The first com­mer­cial ro­bots will reach mar­ket in two to three years at pre­mi­um prices. With­in five to six years, we will have sev­er­al play­ers of­fer­ing mul­ti­ple com­mer­cial ro­bots, some of which will be af­ford­able to busi­ness­es across the Caribbean.

The bind­ing con­straint is not the tech­nol­o­gy it­self but the in­fra­struc­ture to sup­port it. Gath­er­ing train­ing da­ta and sim­u­lat­ing syn­thet­ic da­ta for ro­bot large lan­guage mod­els rep­re­sents the biggest bot­tle­neck. Ear­ly movers will have sig­nif­i­cant ad­van­tages in ro­bot AI, and com­pa­nies like Tes­la al­ready pos­sess­es ob­vi­ous hard­ware ad­van­tages from its man­u­fac­tur­ing ex­per­tise. Nvidia, hav­ing po­si­tioned it­self at the cen­tre of the AI com­put­ing stack, will like­ly lead the open source ro­bot move­ment over time.

For glob­al in­vestors, the im­pli­ca­tions ex­tend far be­yond buy­ing shares in ro­bot­ics com­pa­nies. Brook­field As­set Man­age­ment’s 2026 In­vest­ment Out­look frames the op­por­tu­ni­ty through their “Three Ds” frame­work: dig­i­tal­i­sa­tion, de­glob­al­i­sa­tion and de­car­bon­i­sa­tion. These are not tem­po­rary cy­cles but struc­tur­al trans­for­ma­tions re­shap­ing economies and dri­ving in­vest­ment for decades. The AI and ro­bot­ics rev­o­lu­tion touch­es all three themes si­mul­ta­ne­ous­ly, and T&T sits at an in­ter­est­ing in­ter­sec­tion giv­en our hy­dro­car­bon her­itage and the pres­sures of the glob­al en­er­gy tran­si­tion now bear­ing down up­on us.

Con­sid­er the in­fra­struc­ture re­quire­ments dri­ving glob­al cap­i­tal flows. Glob­al in­fra­struc­ture in­vest­ment needs are ex­pect­ed to ex­ceed US$100 tril­lion by 2040. Ar­ti­fi­cial gen­er­al in­tel­li­gence could un­lock as much as US$10 tril­lion in pro­duc­tiv­i­ty gains over the next decade, but will re­quire US$7 tril­lion of in­fra­struc­ture in­vest­ment across the AI val­ue chain. This in­cludes da­ta cen­tres, ded­i­cat­ed pow­er gen­er­a­tion, com­pute in­fra­struc­ture and semi­con­duc­tor man­u­fac­tur­ing. The num­bers are stag­ger­ing: hy­per­scaler cap­i­tal ex­pen­di­ture is pro­ject­ed to rise from US$270 bil­lion in 2024 to US$1 tril­lion by 2030.

Elec­tric­i­ty is emerg­ing as the crit­i­cal choke­point glob­al­ly. AI work­loads con­sume up to 10 times more pow­er per rack than con­ven­tion­al com­pute, with ex­pec­ta­tions of an­oth­er five to ten­fold in­crease. More than 70 per cent of glob­al trans­mis­sion lines are over 25 years old. An­nu­al grid in­vest­ment will need to ex­ceed US$600 bil­lion by 2030.

In or­der to com­pete, T&T has to en­sure ad­e­quate gen­er­a­tion ca­pac­i­ty in the con­text of the sec­u­lar de­cline in nat­ur­al gas pro­duc­tion. Re­vers­ing the nat­ur­al gas trend is not just for Point Lisas, it is for our abil­i­ty to com­pete as a na­tion.

The emerg­ing mar­kets out­side Chi­na are see­ing some of the fastest growth in elec­tric­i­ty de­mand, yet re­ceive on­ly an es­ti­mat­ed 20 per cent of to­tal an­nu­al in­vest­ment in the pow­er sec­tor. T&T could the­o­ret­i­cal­ly po­si­tion it­self with­in this in­vest­ment flow, but do­ing so re­quires strate­gic think­ing.

The cost of elec­tric­i­ty is ris­ing with the ar­gu­ment that it fa­cil­i­tates nat­ur­al gas ex­ports while at the same time en­er­gy avail­abil­i­ty and costs be­comes a crit­i­cal suc­cess fac­tor in the new era. Our petro­chem­i­cal sec­tor ex­per­tise, our ex­ist­ing en­er­gy in­fra­struc­ture and our ge­o­graph­ic po­si­tion all of­fer la­tent ad­van­tages. But la­tent ad­van­tages re­main per­ma­nent­ly un­re­alised with­out de­lib­er­ate and sus­tained ac­tion from both pub­lic and pri­vate sec­tors.

Labour

The labour mar­ket im­pli­ca­tions de­serve se­ri­ous con­sid­er­a­tion by our pol­i­cy­mak­ers and busi­ness lead­ers alike. Busi­ness­es glob­al­ly may soon pay more for AI agents than peo­ple for cer­tain tasks. This has al­ready hap­pened with con­sumers: In the US, Way­mo rides cost 31 per cent more than Uber on av­er­age, yet de­mand keeps grow­ing be­cause rid­ers pre­fer the safe­ty and re­li­a­bil­i­ty of au­tonomous ve­hi­cles. For rote busi­ness tasks, AIa­gents will com­mand a sim­i­lar pre­mi­um as com­pa­nies fac­tor in on­board­ing, re­cruit­ing, train­ing and man­age­ment costs.

Cur­rent fron­tier AI mod­els re­li­ably com­plete tasks tak­ing peo­ple about an hour. By late 2026, AI agents will au­tonomous­ly ex­e­cute eight-hour plus work­streams. For T&T, where busi­ness process out­sourc­ing and shared ser­vices cen­tres have been tout­ed as di­ver­si­fi­ca­tion strate­gies, this tra­jec­to­ry de­mands ur­gent re­assess­ment.

Job cre­ation in the world of AI and ro­bot­ics will re­quire more na­tion­al at­ten­tion than it is cur­rent­ly re­ceiv­ing. We need to be at the stage of im­ple­ment­ing pro­grams that we have not yet de­signed.

In­dus­tri­al com­pa­nies glob­al­ly are of­ten over­looked and un­der­val­ued, not be­cause their prod­ucts are ob­so­lete but be­cause many have un­der­in­vest­ed in mod­erni­sa­tion and op­er­a­tional ca­pa­bil­i­ties. AI led dig­i­tal­i­sa­tion will trans­form these in­dus­tri­als by cut­ting costs, ad­dress­ing labour short­ages and op­ti­mis­ing sup­ply chains. Our own man­u­fac­tur­ing sec­tor will have to make de­ci­sions around in­vest­ing in au­toma­tion and AI in­te­gra­tion be­cause our re­gion­al and in­ter­na­tion­al com­peti­tors are in that space. Con­sid­er how this will be achieved in the face of a lack of for­eign ex­change avail­abil­i­ty.

Amidst this dis­rup­tion, gen­uine op­por­tu­ni­ties ex­ist for the pre­pared. The com­pa­nies that ben­e­fit from AI-led pro­duc­tiv­i­ty gains will not just be tech­nol­o­gy plat­forms build­ing the mod­els. They will in­clude busi­ness­es in­vest­ing in au­toma­tion and AI tools to ac­cel­er­ate dig­i­tal trans­for­ma­tion. T&T’s fi­nan­cial ser­vices sec­tor, our telecom­mu­ni­ca­tions providers and our more pro­gres­sive man­u­fac­tur­ers could cap­ture sig­nif­i­cant pro­duc­tiv­i­ty gains if they move de­ci­sive­ly. The ques­tion is whether our in­sti­tu­tion­al ca­pac­i­ty, our cap­i­tal mar­kets and our pol­i­cy frame­works can sup­port such moves at the pace re­quired.

The close of 2026 will look dif­fer­ent be­cause the ma­chines are learn­ing to move, to act, to gen­er­ate rather than re­trieve. Glob­al in­vestors who un­der­stand this shift are po­si­tion­ing ac­cord­ing­ly. For this coun­try, the stakes are high­er than mere port­fo­lio re­turns. We are not mere­ly choos­ing in­vest­ments; we are choos­ing whether to re­main rel­e­vant in an econ­o­my be­ing fun­da­men­tal­ly rewired. Those who dis­miss this as dis­tant con­cerns re­lat­ed to the US and Eu­rope, may find them­selves pre­sid­ing over as­sets op­ti­mised for a world that no longer ex­ists. Hap­py New Year 2026 T&T.

Ian Nar­ine is a fi­nan­cial con­sul­tant who ro­bot­i­cal­ly churns out a col­unm every week. Please send your com­ments to ian@ian­nar­ine.com