The con­ver­sa­tion around ar­ti­fi­cial in­tel­li­gence da­ta cen­tres in Trinidad and To­ba­go has been amus­ing, to say the least. It is amus­ing be­cause much of the pop­u­la­tion has jumped straight to the end of the sto­ry with­out pay­ing at­ten­tion to the process re­quired to get there. De­pend­ing on who you lis­ten to, this coun­try is ei­ther about to be­come a cen­tre for ar­ti­fi­cial in­tel­li­gence or is prepar­ing to sur­ren­der its elec­tric­i­ty, wa­ter, land and eco­nom­ic fu­ture to for­eign in­ter­ests. Nei­ther con­clu­sion is sup­port­ed by where the pro­posed projects cur­rent­ly stand.

The pub­lic’s prin­ci­pal in­for­mant has been the me­dia, and sec­tions of the me­dia must ac­cept re­spon­si­bil­i­ty for how the dis­cus­sion has been framed over the past two weeks. Head­lines have too of­ten blurred the dis­tinc­tion be­tween a pro­pos­al, an agree­ment to in­ves­ti­gate a pro­pos­al and an in­vest­ment that is fi­nanced, ap­proved and ready for con­struc­tion. We have even had spu­ri­ous re­port­ing about a se­cret da­ta cen­tre.

For to­day, two points need es­tab­lish­ing. First, what a mem­o­ran­dum of un­der­stand­ing ac­tu­al­ly is. Sec­ond, what eco­nom­ic di­ver­si­fi­ca­tion ac­tu­al­ly means. The pub­lic con­ver­sa­tion has shown lit­tle grasp of ei­ther, which may ex­plain why Trinidad and To­ba­go pro­duces so much talk about di­ver­si­fi­ca­tion and so lit­tle ac­tion.

A mem­o­ran­dum of un­der­stand­ing is a doc­u­ment through which par­ties record their in­ten­tion to ex­plore, pur­sue or co­op­er­ate around a pro­posed ac­tiv­i­ty. It can iden­ti­fy the project, the roles of the par­ties, the in­for­ma­tion they will ex­change and the stud­ies that must be un­der­tak­en be­fore ne­go­ti­a­tions con­tin­ue. It can bring se­ri­ous par­ties to the ta­ble, per­mit con­fi­den­tial in­for­ma­tion to be ex­changed and cre­ate time­lines for due dili­gence. What it can­not do is sub­sti­tute for a com­plet­ed in­vest­ment agree­ment. It is not ev­i­dence that mon­ey has been raised, land trans­ferred, elec­tric­i­ty and wa­ter con­trac­tu­al­ly se­cured, en­vi­ron­men­tal ap­proval grant­ed or a con­struc­tion com­pa­ny ap­point­ed.

The cen­tral com­mer­cial oblig­a­tion, the ac­tu­al com­mit­ment to in­vest and build, typ­i­cal­ly re­mains con­di­tion­al on fur­ther study and de­fin­i­tive agree­ments. An MoU es­tab­lish­es the road along which par­ties in­tend to trav­el. It does not guar­an­tee they reach the des­ti­na­tion.

This dis­tinc­tion is clear­ly record­ed in the two an­nounced pro­pos­als in­volv­ing Ernst & Young and Hum­ming­bird AI Hold­ings. The EY-led pro­pos­al re­port­ed­ly con­tem­plates a 300 megawatt fa­cil­i­ty, with EY in­di­cat­ing it in­tends to work with third par­ties to de­vel­op it. Nei­ther the even­tu­al de­vel­op­er, fi­nancier, op­er­a­tor nor an­chor cus­tomer has been pub­licly iden­ti­fied, so this sits at a very ear­ly stage. The Hum­ming­bird pro­pos­al is some­what more de­fined, con­tem­plat­ing an ini­tial 150 megawatt fa­cil­i­ty with pos­si­ble ex­pan­sion to 500 megawatts, and it has float­ed the first quar­ter of 2028 as an ini­tial tar­get for com­mer­cial op­er­a­tion. A tar­get is not a con­trac­tu­al­ly se­cured com­ple­tion date; it re­mains de­pen­dent on en­gi­neer­ing work, en­vi­ron­men­tal as­sess­ments, util­i­ty analy­sis and reg­u­la­to­ry ap­proval.

In oth­er words it re­mains con­tin­gent on all of the mat­ters that are cur­rent­ly be­ing ob­ject­ed to be­ing ad­e­quate­ly ad­dressed in the pub­lic space. There is through con­struc­tive di­a­logue the abil­i­ty to in­flu­ence the fi­nal in­vest­ment agree­ment, the land agree­ment, elec­tric­i­ty or wa­ter sup­ply con­tracts and the en­vi­ron­men­tal ap­provals.

The same cau­tion ap­plies to the wide­ly cir­cu­lat­ed fig­ure of more than US$5 bil­lion. That num­ber rep­re­sents a po­ten­tial ag­gre­gate val­ue across the two da­ta cen­tre pro­pos­als and a sep­a­rate steel project. It is not mon­ey al­ready in­vest­ed, trans­ferred or con­trac­tu­al­ly com­mit­ted. Po­ten­tial in­vest­ment is not com­mit­ted in­vest­ment; com­mit­ted in­vest­ment is not mon­ey spent; mon­ey spent on stud­ies is not a fi­nan­cial close. Ap­pre­ci­ate that $500 mil­lion was spend on a fea­si­bil­i­ty study for a rapid rail. We do not have a rapid rail in Trinidad and To­ba­go.

Un­der­stand­ing di­ver­si­fi­ca­tion

I have of­ten said in this space that we don’t tru­ly un­der­stand the con­cept of eco­nom­ic di­ver­si­fi­ca­tion. The con­ver­sa­tion around the AI da­ta cen­tres bears that out. Di­ver­si­fi­ca­tion has a pre­cise fi­nan­cial con­struct, best il­lus­trat­ed through a hy­po­thet­i­cal port­fo­lio of com­pa­nies list­ed on the Trinidad and To­ba­go Stock Ex­change.

Be­gin with First Cit­i­zens Fi­nan­cial Hold­ings Lim­it­ed. An in­vestor hold­ing on­ly this stock owns one com­pa­ny and is pri­mar­i­ly ex­posed to in­ter­est rates, cred­it growth and gov­ern­ment fi­nanc­ing needs in Trinidad and To­ba­go. Add Re­pub­lic Bank, whose op­er­a­tions stretch across the Caribbean and Ghana and the com­pa­ny-spe­cif­ic risk falls. But the port­fo­lio still owns two banks, both moved by the same in­ter­est rate and cred­it cy­cles. Add Sco­tia­bank Trinidad and To­ba­go, ma­jor­i­ty owned through the wider Cana­di­an group and you now have a di­ver­si­fi­ca­tion of the own­er­ship of the banks you are in­vest­ed in. How­ev­er, its lo­cal earn­ings re­main tied to the do­mes­tic econ­o­my. You are in­vest­ed in three dif­fer­ent com­pa­nies but still ex­posed to one sec­tor, fi­nan­cial ser­vices.

You can’t speak about di­ver­si­fi­ca­tion with­out un­der­stand­ing cor­re­la­tion. Cor­re­la­tion mea­sures how close­ly two in­vest­ments move to­geth­er; a co­ef­fi­cient near one means they rise and fall in tan­dem, near ze­ro means lit­tle con­sis­tent re­la­tion­ship. Own­ing three banks re­duces the risk that one in­sti­tu­tion’s fail­ure sinks the port­fo­lio, but it does noth­ing to pro­tect against a shock to bank­ing as a whole.

Now, add Guardian Hold­ings and the port­fo­lio gains an in­sur­ance el­e­ment. You are still in fi­nan­cial ser­vices but now it’s in­sur­ance and bank­ing. Add Massy Hold­ings and you gain re­tail and dis­tri­b­u­tion earn­ings, dri­ven by con­sumer spend­ing and sup­ply chains across sev­er­al ter­ri­to­ries. Add ANSA McAL and you gain man­u­fac­tur­ing, au­to­mo­tive and bev­er­age in­ter­ests span­ning mul­ti­ple mar­kets. Add Trinidad and To­ba­go NGL, whose earn­ings flow chiefly through its stake in Phoenix Park Gas Proces­sors and your port­fo­lio ac­quires en­er­gy sec­tor ex­po­sure.

It would be im­pre­cise to claim the mea­sured cor­re­la­tion must fall with every ad­di­tion; the re­sult de­pends on weight­ings and the pe­ri­od ex­am­ined, but the prin­ci­ple holds. As gen­uine­ly dif­fer­ent in­come sources are added, the port­fo­lio de­pends less on any sin­gle com­pa­ny, sec­tor or eco­nom­ic fac­tor. The aim was nev­er to own the largest num­ber of names. It was to stop own­ing the same risk re­peat­ed­ly in dif­fer­ent com­pa­nies.

A coun­try’s econ­o­my works the same way. Trinidad and To­ba­go has pro­duced oil, nat­ur­al gas, LNG, am­mo­nia, methanol and nat­ur­al gas liq­uids, dif­fer­ent prod­ucts from dif­fer­ent plants sold to dif­fer­ent cus­tomers, yet all sub­stan­tial­ly con­nect­ed to glob­al en­er­gy prices, gas avail­abil­i­ty and in­dus­tri­al de­mand. That re­sem­bles own­ing First Cit­i­zens, Re­pub­lic and Sco­tia­bank to­geth­er. We may hold sev­er­al se­cu­ri­ties while ef­fec­tive­ly hold­ing one trade.

True di­ver­si­fi­ca­tion does not re­quire aban­don­ing en­er­gy. A sen­si­ble in­vestor keeps a pro­duc­tive as­set and us­es its re­turns to ac­quire dif­fer­ent risk. T&T’s en­er­gy re­sources and in­dus­tri­al ca­pa­bil­i­ty should serve as the plat­form from which new, dif­fer­ent­ly dri­ven in­come is built. Mov­ing from crude oil in­to LNG, am­mo­nia and methanol added pro­cess­ing and ex­port val­ue, but it re­mained cor­re­lat­ed with en­er­gy eco­nom­ics. It was di­ver­si­fi­ca­tion with­in the en­er­gy port­fo­lio rather than away from it.

Now there is an op­por­tu­ni­ty for some­thing dif­fer­ent. An AI da­ta cen­tre’s rev­enue would come from de­mand for com­put­ing ca­pac­i­ty and cloud ser­vices, not from am­mo­nia or methanol prices. A fall in gas prices does not au­to­mat­i­cal­ly re­duce glob­al de­mand for com­pu­ta­tion, which cre­ates a gen­uine­ly less cor­re­lat­ed in­come stream, re­sem­bling the ad­di­tion of ANSA McAL or Massy to a bank heavy port­fo­lio. But a large fa­cil­i­ty still needs sub­stan­tial elec­tric­i­ty, and if that pow­er comes from nat­ur­al gas, an im­por­tant link to the en­er­gy sec­tor sur­vives. The dif­fer­ence lies in whether gas is an in­put in­to a high­er val­ue ser­vice or the fi­nal prod­uct it­self sold at a com­mod­i­ty price.

We are not yet able to ad­ju­di­cate on that be­cause, as I in­di­cat­ed ear­li­er,there are de­tails that are still out­stand­ing. What I can say is that we maybe head­ing in the right di­rec­tion. We do need to know who owns the fa­cil­i­ty, who sup­plies the cap­i­tal, who buys its out­put, how much for­eign ex­change stays here, what is paid for elec­tric­i­ty and wa­ter, how many lo­cal firms and work­ers par­tic­i­pate, and what risks the State is asked to car­ry. A fa­cil­i­ty run­ning on sub­sidised pow­er, im­port­ing its equip­ment and labour and repa­tri­at­ing its rev­enue adds out­put with­out di­ver­si­fy­ing in­come. One that earns for­eign ex­change com­mer­cial­ly, trains lo­cal work­ers and an­chors a wider dig­i­tal ecosys­tem is a dif­fer­ent propo­si­tion en­tire­ly.

The MoUs signed so far can­not yet an­swer these ques­tions. That is ex­act­ly why there must be trans­paren­cy around due dili­gence, en­vi­ron­men­tal analy­sis, util­i­ty ne­go­ti­a­tion etc. It is al­so why the pub­lic should en­gage con­struc­tive­ly as op­posed to the cur­rent ap­proach. We do need to di­ver­si­fy our econ­o­my at scale. What we have done up un­til now isn’t suf­fi­cient. We now have a re­al di­ver­si­fi­ca­tion op­por­tu­ni­ty to con­sid­er.

It would be reck­less to dis­miss the pro­pos­al with­out ex­plor­ing the fea­si­bil­i­ty and work­ing to en­sure that a re­al ben­e­fit ac­crues to Trinidad and To­ba­go.

Ian Nar­ine is a fi­nan­cial con­sul­tant who is hop­ing to see more nat­ur­al in­tel­li­gence. Please send your com­ment to ian@ian­nar­ine.com