AI Capex Accelerates, Supporting Bloom Energy and Manhattan Associates

By Daniel Okafor

FactSet consensus puts SpaceX’s 2027 capital spending at $249.42 billion, versus an estimated $80.4 billion this year, according to MarketWatch’s review of analyst forecasts.

MarketWatch framed Sen’s two stock choices as ways to gain exposure to infrastructure and enterprise software as AI spending expands.

Sowmdeb Sen, a senior portfolio manager at Silvant Capital Management in Atlanta, said during a MarketWatch interview: “The key metrics we monitor for AI are actually accelerating.” He co-manages the Virtus Silvant Mid-Cap Growth Fund, which Morningstar gives five stars in the “Mid-Cap Growth” category, its highest rating.

What is driving the spending cycle?

MarketWatch said rising rates can pull money out of stocks even when the economy performs well, so long-term investors should look beyond daily headlines. Sen still sees attractive opportunities after months of turmoil in AI shares.

MarketWatch reported that the yield on 10-year U.S. Treasury notes hit a 2026 intraday peak of 5.37% on Oct. 7, up from 4.17% at the end of last year. It eased to 4.97% early Monday. Bond prices move opposite yields.

MarketWatch said higher yields raise borrowing costs for hyperscalers and other companies funding AI infrastructure. Sen said spending on data centers, related equipment and software keeps increasing as companies and consumers find more uses for AI. He considers the market still early in the spending cycle.

How far can AI capital spending run?

SpaceX CEO Elon Musk said on an Oct. 20 earnings call that he was fairly confident the company could bring an incremental 10 gigawatts of compute online in 2027. An LSEG transcript put expected year-end online capacity at 2.5 to 3 gigawatts.

Musk also said: “We’re actually going to aim for an incremental 20 gigawatts of power and cooling infrastructure next year.” He cautioned that execution could fall short, but said the plan would put total capacity at roughly 23 gigawatts.

SpaceX CFO Bret Johnsen said fourth-quarter capital expenditures would range from $35 billion to $40 billion, above reported third-quarter capex of $19.59 billion. Sen expected SpaceX’s annual capital spending to reach $200 billion, below FactSet’s 2027 consensus.

Meta Platforms said on Monday it planned to raise at least $95 billion through stock offerings. Meta expects 2026 capex of $135 billion to $145 billion, while FactSet consensus puts 2027 spending at $208.04 billion. MarketWatch also reported that Nvidia, AMD and Broadcom could buy Meta shares as part of a $35 billion stock offering.

Why Bloom Energy?

Sen picked Bloom Energy for the data-center construction wave. The company builds on-site power-generation equipment fueled by natural gas, biogas and hydrogen. A new data center can wait two years for a traditional grid connection, Sen said.

Sen said Bloom can create energy from hydrogen gas in three to four months. He said that deployment speed could matter when developers cannot wait for utilities.

On Thursday, Bloom reported third-quarter revenue of $1.24 billion, up 138% from the third quarter of 2025. FactSet consensus expects 2027 sales to rise 63% to $7.16 billion, from an expected $4.4 billion in 2026. The 2027 estimate has increased 12% since the end of June.

MarketWatch, using consensus 12-month estimates, put Bloom’s valuation above the S&P 500’s:

• Forward P/E: 39.6 times, versus the index’s weighted 18.7.
• Forward price-to-sales: 8.6 times, versus 3.2.
• Revenue CAGR, 2026-2028: 54%, versus 8.6%.
• EPS CAGR, 2026-2028: 65.4%, versus 16.4%.

MarketWatch framed that premium as the price of rapid growth. Bloom’s valuation case depends on FactSet’s forecasts arriving, while the 12% estimate revision provides the latest support for demand.

Why Manhattan Associates?

Sen’s second pick, Manhattan Associates, develops supply-chain software for retailers. Investors had written the company off last year amid fears that AI would displace its business, Sen said. The stock fell 34% then but has risen 37.6% so far in 2026.

Sen said Silvant’s discussions with Manhattan management, customer surveys and conference comments indicated that retailers were using more Manhattan AI products rather than OpenAI or customized tools. Some customers reported 15% cost declines after introducing Manhattan AI agents, he added.

Institutional investors had expected Manhattan’s cloud-services growth to slow, and consensus estimates reflected that view. In the second quarter, Manhattan reported 26% year-over-year cloud revenue growth and 9% total revenue growth.

MarketWatch put Manhattan’s valuation at:

• Forward P/E: 39.6 times.
• Forward price-to-sales: 11.1 times.
• Consensus sales CAGR, 2026-2028: 8.8%.
• Consensus EPS CAGR, 2026-2028: 13.5%.

Sen said his team looks for companies that can beat analysts, investors, management and competitors. Manhattan’s cloud growth met that test. Bloom’s rising estimate offers a parallel test for its power-equipment business.

What could validate the thesis?

FactSet’s Bloom forecast sets the next test: investors need delivery on 63% 2027 sales growth and further estimate revisions. For Manhattan, the test is whether cloud growth keeps exceeding the slower expectations embedded in consensus.

MarketWatch’s valuation data put both stocks at 39.6 times forward earnings. The cited evidence is rising capital spending, Bloom’s estimate revisions and Manhattan’s cloud growth. Sen’s thesis depends on those operating results, not on a broad rebound in AI sentiment or lower interest rates.