VGT Can Keep Winning While AI Still Has Time to Prove Itself

VGT's case does not depend on a clean AI verdict

The core opportunity is straightforward: investors do not need a final AI verdict to keep owning VGT, though waiting too long could mean paying up for the next leg of leadership. VGT has compounded about 14% annualized return since 2004, manages $147.3 billion in net assets, charges a 0.09% expense ratio, and has 60.59% of assets in its top 10 holdings. That last point is the real tension: the fund's strength and its risk come from the same place, a heavy tilt toward the megacaps and semiconductor leaders investors already associate with the AI wave.

That setup invites both discipline and bias. In 2025, VGT gained 23% versus 17% for the S&P 500. Bulls can argue that the market is still rewarding exposure to software, semiconductors, and cloud-linked infrastructure. Bears can argue that recent outperformance is being mistaken for proof that leadership will simply continue.

What matters over the next few quarters is less the rhetoric around AI than the underlying flow of spending and revenue. If infrastructure spending holds and cloud providers keep monetizing demand, VGT can continue to outperform even while the broader AI debate stays messy. The risk is straightforward too: premium valuations and lighter diversification can make drawdowns sharper if spending cools or leadership broadens.

VGT already spans the parts of AI getting paid first

The advantage here is timing within the stack, not a clean declaration that AI has "arrived."

In the current AI buildout, spending tends to move in stages: first chips, then the infrastructure that moves and stores data, and later the software layer that turns compute into recurring revenue. VGT already has exposure across that chain. Its top holdings span Nvidia at 16.78%, Apple at 15.26%, and Microsoft at 9.87%, with additional semicap AI infrastructure and semiconductor exposure through Broadcom at 4.49%, Micron at 4.19%, AMD at 3.20%, and Lam Research at 1.55%. That means VGT does not need the long-term software payoff to be fully proven for some AI beneficiaries to keep posting results.

Why the sequence matters

This is why the fund's mix matters more than its concentration at this stage. AI data-center demand typically starts with GPUs, then networking, memory, and manufacturing equipment before enterprise software monetization is fully visible in the aggregate. VGT already owns part of that picks-and-shovels layer, so it can benefit if spending stays strong even if the market still needs time to validate the broader software narrative.

The practical point is simple: - VGT can do well if AI spending keeps supporting the hardware and infrastructure layer. - It is not entirely dependent on the later software monetization story arriving all at once. - The watchpoint is whether spending cools before monetization broadens.

The risk investors may be underestimating is concentration

VGT's track record can make concentration feel like a feature rather than a risk. A fund that has gained over 1,800% since inception and posted a 23% gain in 2025 versus 17% for the S&P 500 creates powerful anchoring. Once investors get used to tech leadership enduring, normal pullbacks can start to feel like exceptions.

The structural issue is not quality; it is exposure. With 60.59% of assets in the top 10 holdings, VGT is not a broad diversification shield. It is a concentrated bet that a relatively small group of giants can keep compounding fast enough to carry the fund. If AI infrastructure spending continues to concentrate in current leaders, that structure can keep working. If not, the same concentration can amplify a reversal.

What could weaken the thesis

The first crack may come from valuation compression and uneven earnings rather than a lack of innovation. After a period of clear outperformance, that is often when concentration stops being the edge and becomes the constraint.

Watch for these invalidation conditions: - AI demand remains narrow, with monetization failing to broaden beyond a small core. - Infrastructure spending cools before service providers can turn demand into broader profits. - Market leadership broadens away from tech, raising the odds that the S&P 500 outperforms again.

What to monitor if you are staying with the monetization trail

The cleaner way to assess VGT here is not to argue about AI in the abstract, but to follow where spending turns into revenue. After a 23% gain in 2025 versus 17% for the S&P 500, it is easy to treat excitement as proof. A more disciplined filter is simpler: watch whether the money first pays the chip, network, memory, and cloud builders, and only later reaches the softer software stories.

Bullish signposts

• Cloud and infrastructure names keep converting demand into revenue.
• Corporate AI spending remains firm across GPUs, memory, networking, and data centers.
• Leadership stays concentrated in the companies already capturing that cash.

Invalidation signals

• Earnings confirm AI demand, but monetization stays narrow.
• Spending cools before service providers can turn backlog into profits.
• Market leadership broadens away from tech, increasing the odds that the S&P 500 outperforms.

VGT does not need AI to be perfectly proven right now. It needs the market to keep paying the builders first. If that order changes, the thesis changes with it.