MIT’s gonorrhea treatment revives hope that AI isn’t just Fool’s Code

(Originally published Aug. 15 in “What in the World“) Maybe this AI stuff isn’t all hocus-pocus after all.

Researchers at the Massachusetts Institute of Technology, a sanctum of America’s fast-declining real intelligence under Trump, report that they used artificial intelligence to design potential antibiotics against drug-resistant gonorrhea and MRSA (methicillin-resistant Staphylococcus aureus). Leaving aside why MIT’s researchers chose these two particular, contact-based infections to sick their bot on, the MIT team trained their model to comb through billions of chemical structures, flagging those most likely to work on these cooties.

This should give pause to those of us who were just about to write AI off as just another tech-industry slogan for ever-more powerful computing—the successor to “the network is the computer,” “the Cloud,” “software-as-a-service,” and “robotic process automation.”

MIT’s breakthrough is part of the glimmer of evidence that AI is finally starting to yield more than delusions of grandeur and animated baby Trump shorts. Aside from curing the clap, MIT appears to have demonstrated a practical way AI can— and this is profoundly hum-drum—outsource and accelerate the drudgery of research to machines.

It’s still just a glimmer. Despite pouring billions into developing and using AI, 80% of companies have yet to discern any real impact. Yet from customer service to operational planning, from college labs to boardrooms and trading floors, AI is definitely changing the way we do, well, everything.

AI’s more tangible impact, of course, is not what it has yielded in productivity. Oh no. Its most amazing achievement, the one that truly boggles the mind, is just how much money AI has managed to suck in without having much demonstrably useful output at all. IDC estimates businesses will double spending this year on generative AI, to $62 billion.

Will it work? Who cares? Just building the data centers capable of handling the massive computing tasks labeled “AI” has fueled a massive boom in assembling computer servers, the power to run them, and the land to put them on. Each of Big Tech’s mammoth AI data centers—from Meta’s “Prometheus” and “Hyperion,” to OpenAI’s “Stargate” and Elon Musk’s xAI “Colossus” represent investments of more than $100 billion.

Total global spending on AI-related infrastructure, analysts at Morgan Stanley estimate, could reach $3 trillion by 2029. Roughly half of that will come from Big Tech’s massive cash piles. The rest will need to be financed by bankers, hedge funds, investors, private equity firms, and sovereign wealth funds. The scale of funds needed is creating its own boomlet in finance—from asset-backed securities and construction loans, to green bonds and project finance.

The AI Gold Rush may turn up a dry hole, but manufacturing is shifting fast to sell picks and shovels to its prospectors: in the second quarter, Apple’s largest contract electronics supplier, Foxconn, sold more AI servers than smartphones for the first time ever. Cloud and networking products now account for over 40% of its revenue, and the company expects AI server sales to triple in the next quarter.

Skepticism still abounds about AI. Many still think it’ll turn out to be a bunch of pyrite. But the money pouring in is starting to turn Wall Street’s most skeptical heads. Hedge funds including Bridgewater Associates, Discovery Capital, and Tiger Global have all boosted their exposure to AI-driven Big Tech—Google, Meta, Microsoft, and Nvidia.

Momentum in markets begets a feeding frenzy: the more capital flows into AI, the faster the ecosystem grows, the more attractive it becomes for both established players and newcomers, and the more absurdly large the bets grow. Wit hedge funds like Coatue and Discovery throwing money into crytpo-bugs turned data-center convert CoreWeave have driven its market capitalization to $49 billion.

AI is delivering real benefits to some. Staffing and talent firms such as Adecco, Kelly Services, ManpowerGroup, Robert Half, and Randstad are all developing AI platforms to automate candidate screening, optimize workforce planning, and improve operational efficiency.

Yet AI’s fantastic growth has already given it a distinctively bubblish sheen. Risks of overcapacity and obsolescence abound. Many new data centers are being build using hardware—and even financed with that hardware as collateral—that is likely to be obsolete before the ribbon is cut. That’s drawing comparisons to the telecom bubble of the late-1990s, when companies laid too many cables and projected perpetually rising demand, only to see it all crumble.

But California wasn’t built on gold. It was built, as Foxconn is re-discovering, on selling goods and services around the search for gold. So, whether or not AI delivers on its promise, its sheer size and speed of the AI juggernaut are making it the latest unstoppable economic force. Ignoring AI is no longer an option. And at least for now, it is being treated like the wheel, the printing press, electricity, the telephone, and the computer, as one of those advances that re-writes not just industry, but history.

That, or it could go down as that burning sensation that followed that night of too many cocktails at that MIT fraternity party that time.