Wells Fargo’s Ohsung Kwon: 2027 will be a year of AI applications over AI buildout

LITTLE BIT TODAY WHEN IT COMES TO YIELD, BUT THEY'RE DOWN A LITTLE BIT FROM WHERE THEY WERE EARLIER THIS MORNING. >> STEVE, THANKS FOR THAT. STEVE LIESMAN ON NEW YORK FED THIS MORNING. FOR MORE ON HOW THE MARKETS ARE DIGESTING SOME OF THOSE NUMBERS, LET'S BRING IN WELLS FARGO. CHIEF EQUITY STRATEGIST. HONG KONG JOINS US HERE AT POST NINE. GREAT TO HAVE YOU BACK. THANKS FOR HAVING ME. WE WERE JOKING OFF CAMERA DURING THE SUMMER. YOU DID SAY SELL WORLD CUP. AND THEN YOU GOT MORE CAUTIOUS IN SEPTEMBER. WHERE DOES IT WHERE DOES IT LEAVE YOUR HEAD RIGHT NOW GOING INTO YOUR END? YEAH, I THINK IT'S JUST GOING TO BE CHOPPY THROUGH THE YEAR END. AND WE'RE STARTING TO GET A LITTLE MORE CONCERNED ABOUT THE AI CAPEX CYCLE BEYOND 2027. SO WE TRIED TO ESTIMATE HOW BIG AI HAS TO GET TO JUSTIFY CAPEX THROUGH 27. AND THE OVERALL AI DEMAND HAS TO GET TO ABOUT $1.4 TRILLION BY 28. AND THAT'S ABOUT $9,000 PER WORKER PER YEAR. SO THAT'S A SIGNIFICANT THAT'S A PRETTY SIGNIFICANT NUMBER. AND IF HYPERSCALERS WERE TO DO ANOTHER BIG CAPEX IN 2028, THAT NUMBER JUMPS TO $2.2 TRILLION. SO THE BOTTOM LINE IS EVERY YEAR IT BECOMES INCREASINGLY MORE DIFFICULT FOR HYPERSCALERS TO JUSTIFY ANOTHER BIG CAPEX BEYOND 27. SO I REALLY THINK THE RISK IS THAT CAPEX CYCLE IS STARTING TO NOT ROLL OVER COMPLETELY, BUT FLAT LINES AFTER 27. AND I DON'T THINK THE BUILD OUT TRADE IS NECESSARILY PRICING IN THAT RISK YET. >> WELL, WE'RE ESSENTIALLY WHEN YOU LOOK AT CONSENSUS EXPECTATION ON HYPERSCALER CAPEX, WE'RE KIND OF AT THE PEAK IN TERMS OF YEAR ON YEAR GROWTH, RIGHT? I MEAN, CONSENSUS HAS BEEN HABITUALLY WRONG ON THE DOWNSIDE, BUT THIS IS WHERE THEY THINK IT ESSENTIALLY TOPS OUT ON GROWTH. >> YEAH. SO THE RATE OF CHANGE ARGUMENT, THE RATE OF CHANGE IS ESSENTIALLY GOING TO PEAK. NOW HYPERSCALE CAPEX AS WELL AS SEMI EARNINGS. SO THE RATE OF CHANGE ARGUMENT IS GOING AGAINST SEMIS BASICALLY. AND WE'RE STARTING TO LIKE HYPERSCALERS BECAUSE OF THAT REASON. BECAUSE THE RATE OF CHANGE ON CAPEX IS SLOWING DOWN. WHEREAS THE REVENUE SIDE IS STARTING TO ACCELERATE. SO I THINK THE FREE CASH FLOW ARGUMENT, THE BEAR CASE FOR HYPERSCALERS THAT'S WEAKENING INTO 2027, AND I KNOW IT'S A LITTLE TOO EARLY, BUT IF I HAVE TO START THINKING ABOUT 2027, I THINK IT COULD ACTUALLY BE A YEAR WHERE AI APPLICATIONS WIN OVER THE AI BUILD OUT, MEANING HYPERSCALERS INFRASTRUCTURE, SOFTWARE OVER SEMIS. >> SO LESS PICKS AND SHOVELS AND START BUYING THE ROI FROM THE USERS. YEAH. ESSENTIALLY IT. >> YEAH. >> BUT EVEN EVEN IN A PERIOD, IF WE CONTINUE TO SEE THESE RISING RATES, AS I SAID, I KNOW YOU YOU SAID THE CASH FLOW CONCERNS WEREN'T AS PROBLEMATIC. BUT YOU KNOW, THE DEBT BINGE IS REAL. >> YEAH. I MEAN, RATES ARE CONCERNING. I DON'T THINK IT WILL REALLY HAVE THAT FUNDAMENTAL IMPACT ON HYPERSCALERS OR SEMIS. YET. WHAT I'M MORE CONCERNED ABOUT IS THE VALUATION ASPECT ASPECT OF IT. SO A LOT OF PEOPLE SAY THE S&P 500 HAS DERATED ENOUGH. AND YOU KNOW, IF YOU LOOK AT THE FOUR MULTIPLES, IT'S TRADING AT ABOUT 19 TIMES RIGHT NOW VERSUS THE HIGHS OF 23. BUT THE ENTIRE DERATING HAS ACTUALLY BEEN DRIVEN BY HIGHER RATES. AND IF YOU ADJUST FOR HIGHER RATES, THE EQUITY RISK PREMIUM HAS ACTUALLY COMPRESSED QUITE A LOT. AND RIGHT NOW IT'S ACTUALLY NEGATIVE. SO IF YOU JUST LOOK AT THE FORWARD EARNINGS YIELD VERSUS THE TEN YEAR AT 530 RIGHT NOW IT'S ABOUT NEGATIVE TEN BASIS POINTS. AND IT'S AT THE CYCLE LOW. THE ONLY TIME IT WENT NEGATIVE HISTORICALLY HAS BEEN DURING THE TECH BUBBLE. SO IT'S I THINK IT'S GOING TO BE HARD TO ARGUE THAT THE EQUITY RISK PREMIUM IS GOING TO COMPRESS EVEN FURTHER FROM HERE. AND THAT'S BEEN REALLY THE OFFSETTING FACTOR OF HIGHER RATES, MEANING IF RATES WERE TO CLIMB HIGHER FROM HERE, THERE'S NOT GOING TO BE THAT OFFSETTING FACTOR FROM THE EQUITY RISK PREMIUM. SO I THINK HIGHER RATES FROM HERE IS GOING TO BE A LOT MORE PAINFUL GOING FORWARD. >> AND YOU ACTUALLY PUT SOME NUMBERS ON HOW MANY BASIS POINTS IN THE INCREASE IN R